id sid tid token lemma pos ajst-10994 1 1 academic academic ADJ ajst-10994 1 2 journal journal NOUN ajst-10994 1 3 of of ADP ajst-10994 1 4 science science NOUN ajst-10994 1 5 and and CCONJ ajst-10994 1 6 technology technology NOUN ajst-10994 1 7 issn issn NOUN ajst-10994 1 8 : : PUNCT ajst-10994 1 9 2771 2771 NUM ajst-10994 1 10 - - SYM ajst-10994 1 11 3032 3032 NUM ajst-10994 1 12 | | NOUN ajst-10994 1 13 vol vol NOUN ajst-10994 1 14 . . PROPN ajst-10994 2 1 7 7 NUM ajst-10994 2 2 , , PUNCT ajst-10994 2 3 no no INTJ ajst-10994 2 4 . . NOUN ajst-10994 2 5 1 1 NUM ajst-10994 2 6 , , PUNCT ajst-10994 2 7 2023 2023 NUM ajst-10994 2 8 79 79 NUM ajst-10994 2 9 research research NOUN ajst-10994 2 10 on on ADP ajst-10994 2 11 fault fault NOUN ajst-10994 2 12 diagnosis diagnosis NOUN ajst-10994 2 13 method method NOUN ajst-10994 2 14 of of ADP ajst-10994 2 15 rotary rotary ADJ ajst-10994 2 16 machinery machinery NOUN ajst-10994 2 17 based base VERB ajst-10994 2 18 on on ADP ajst-10994 2 19 improved improved ADJ ajst-10994 2 20 transformer transformer ADJ ajst-10994 2 21 haijie haijie NOUN ajst-10994 2 22 zhi1 zhi1 PROPN ajst-10994 2 23 , , PUNCT ajst-10994 2 24 * * PUNCT ajst-10994 2 25 , , PUNCT ajst-10994 2 26 jinkui jinkui PROPN ajst-10994 2 27 wang1 wang1 PROPN ajst-10994 2 28 , , PUNCT ajst-10994 2 29 a a DET ajst-10994 2 30 , , PUNCT ajst-10994 2 31 haitao haitao PROPN ajst-10994 2 32 zhang1 zhang1 PROPN ajst-10994 2 33 , , PUNCT ajst-10994 2 34 2 2 NUM ajst-10994 2 35 , , PUNCT ajst-10994 2 36 a a PRON ajst-10994 2 37 , , PUNCT ajst-10994 2 38 yongkang yongkang PROPN ajst-10994 2 39 hou1 hou1 PROPN ajst-10994 2 40 , , PUNCT ajst-10994 2 41 a a DET ajst-10994 2 42 , , PUNCT ajst-10994 2 43 qishun qishun PROPN ajst-10994 2 44 yang1 yang1 PROPN ajst-10994 2 45 , , PUNCT ajst-10994 2 46 a a DET ajst-10994 2 47 1 1 NUM ajst-10994 2 48 college college NOUN ajst-10994 2 49 of of ADP ajst-10994 2 50 engineering engineering PROPN ajst-10994 2 51 , , PUNCT ajst-10994 2 52 china china PROPN ajst-10994 2 53 university university PROPN ajst-10994 2 54 of of ADP ajst-10994 2 55 petroleum petroleum NOUN ajst-10994 2 56 ( ( PUNCT ajst-10994 2 57 beijing beijing PROPN ajst-10994 2 58 ) ) PUNCT ajst-10994 2 59 karamay karamay PROPN ajst-10994 2 60 campus campus PROPN ajst-10994 2 61 , , PUNCT ajst-10994 2 62 karamay karamay PROPN ajst-10994 2 63 , , PUNCT ajst-10994 2 64 834000 834000 NUM ajst-10994 2 65 , , PUNCT ajst-10994 2 66 china china PROPN ajst-10994 2 67 2 2 NUM ajst-10994 2 68 college college NOUN ajst-10994 2 69 of of ADP ajst-10994 2 70 software software NOUN ajst-10994 2 71 , , PUNCT ajst-10994 2 72 taiyuan taiyuan PROPN ajst-10994 2 73 university university PROPN ajst-10994 2 74 of of ADP ajst-10994 2 75 technology technology PROPN ajst-10994 2 76 , , PUNCT ajst-10994 2 77 taiyuan taiyuan PROPN ajst-10994 2 78 , , PUNCT ajst-10994 2 79 030024 030024 NUM ajst-10994 2 80 , , PUNCT ajst-10994 2 81 china china PROPN ajst-10994 2 82 athe athe PROPN ajst-10994 2 83 contributions contribution NOUN ajst-10994 2 84 of of ADP ajst-10994 2 85 these these DET ajst-10994 2 86 authors author NOUN ajst-10994 2 87 to to ADP ajst-10994 2 88 this this DET ajst-10994 2 89 paper paper NOUN ajst-10994 2 90 are be AUX ajst-10994 2 91 consistent consistent ADJ ajst-10994 2 92 * * PUNCT ajst-10994 2 93 corresponding correspond VERB ajst-10994 2 94 author author NOUN ajst-10994 2 95 : : PUNCT ajst-10994 2 96 haijie haijie PROPN ajst-10994 2 97 zhi zhi PROPN ajst-10994 2 98 ( ( PUNCT ajst-10994 2 99 email email NOUN ajst-10994 2 100 : : PUNCT ajst-10994 2 101 1810365936@qq.com 1810365936@qq.com NUM ajst-10994 2 102 ) ) PUNCT ajst-10994 2 103 abstract abstract NOUN ajst-10994 2 104 : : PUNCT ajst-10994 2 105 in in ADP ajst-10994 2 106 modern modern ADJ ajst-10994 2 107 industry industry NOUN ajst-10994 2 108 , , PUNCT ajst-10994 2 109 rotating rotate VERB ajst-10994 2 110 machinery machinery NOUN ajst-10994 2 111 plays play VERB ajst-10994 2 112 a a DET ajst-10994 2 113 crucial crucial ADJ ajst-10994 2 114 role role NOUN ajst-10994 2 115 . . PUNCT ajst-10994 3 1 these these DET ajst-10994 3 2 rotating rotate VERB ajst-10994 3 3 machines machine NOUN ajst-10994 3 4 are be AUX ajst-10994 3 5 not not PART ajst-10994 3 6 only only ADV ajst-10994 3 7 fundamental fundamental ADJ ajst-10994 3 8 components component NOUN ajst-10994 3 9 of of ADP ajst-10994 3 10 power power NOUN ajst-10994 3 11 generation generation NOUN ajst-10994 3 12 and and CCONJ ajst-10994 3 13 propulsion propulsion NOUN ajst-10994 3 14 systems system NOUN ajst-10994 3 15 but but CCONJ ajst-10994 3 16 also also ADV ajst-10994 3 17 key key ADJ ajst-10994 3 18 factors factor NOUN ajst-10994 3 19 for for ADP ajst-10994 3 20 their their PRON ajst-10994 3 21 efficient efficient ADJ ajst-10994 3 22 operation operation NOUN ajst-10994 3 23 . . PUNCT ajst-10994 4 1 harsh harsh ADJ ajst-10994 4 2 operating operating NOUN ajst-10994 4 3 environments environment NOUN ajst-10994 4 4 often often ADV ajst-10994 4 5 lead lead VERB ajst-10994 4 6 to to ADP ajst-10994 4 7 failures failure NOUN ajst-10994 4 8 in in ADP ajst-10994 4 9 critical critical ADJ ajst-10994 4 10 components component NOUN ajst-10994 4 11 like like ADP ajst-10994 4 12 gears gear NOUN ajst-10994 4 13 and and CCONJ ajst-10994 4 14 bearings bearing NOUN ajst-10994 4 15 in in ADP ajst-10994 4 16 rotating rotate VERB ajst-10994 4 17 machinery machinery NOUN ajst-10994 4 18 , , PUNCT ajst-10994 4 19 which which PRON ajst-10994 4 20 can can AUX ajst-10994 4 21 directly directly ADV ajst-10994 4 22 result result VERB ajst-10994 4 23 in in ADP ajst-10994 4 24 equipment equipment NOUN ajst-10994 4 25 malfunction malfunction NOUN ajst-10994 4 26 . . PUNCT ajst-10994 5 1 therefore therefore ADV ajst-10994 5 2 , , PUNCT ajst-10994 5 3 authentic authentic ADJ ajst-10994 5 4 fault fault NOUN ajst-10994 5 5 diagnosis diagnosis NOUN ajst-10994 5 6 of of ADP ajst-10994 5 7 these these DET ajst-10994 5 8 primary primary ADJ ajst-10994 5 9 building building NOUN ajst-10994 5 10 blocks block NOUN ajst-10994 5 11 in in ADP ajst-10994 5 12 rotating rotate VERB ajst-10994 5 13 machinery machinery NOUN ajst-10994 5 14 is be AUX ajst-10994 5 15 of of ADP ajst-10994 5 16 pivotal pivotal ADJ ajst-10994 5 17 value value NOUN ajst-10994 5 18 for for ADP ajst-10994 5 19 improving improve VERB ajst-10994 5 20 its its PRON ajst-10994 5 21 reliability reliability NOUN ajst-10994 5 22 and and CCONJ ajst-10994 5 23 safety safety NOUN ajst-10994 5 24 during during ADP ajst-10994 5 25 operation operation NOUN ajst-10994 5 26 . . PUNCT ajst-10994 6 1 this this DET ajst-10994 6 2 research research NOUN ajst-10994 6 3 proposes propose VERB ajst-10994 6 4 a a DET ajst-10994 6 5 fault fault NOUN ajst-10994 6 6 diagnosis diagnosis NOUN ajst-10994 6 7 strategy strategy NOUN ajst-10994 6 8 for for ADP ajst-10994 6 9 rotating rotate VERB ajst-10994 6 10 machinery machinery NOUN ajst-10994 6 11 rooted root VERB ajst-10994 6 12 on on ADP ajst-10994 6 13 time time NOUN ajst-10994 6 14 series series NOUN ajst-10994 6 15 transformer transformer NOUN ajst-10994 6 16 and and CCONJ ajst-10994 6 17 validates validate VERB ajst-10994 6 18 the the DET ajst-10994 6 19 effectiveness effectiveness NOUN ajst-10994 6 20 of of ADP ajst-10994 6 21 the the DET ajst-10994 6 22 proposed propose VERB ajst-10994 6 23 approach approach NOUN ajst-10994 6 24 . . PUNCT ajst-10994 7 1 firstly firstly ADV ajst-10994 7 2 , , PUNCT ajst-10994 7 3 a a DET ajst-10994 7 4 time time NOUN ajst-10994 7 5 series series NOUN ajst-10994 7 6 transformer transformer NOUN ajst-10994 7 7 model model NOUN ajst-10994 7 8 is be AUX ajst-10994 7 9 designed design VERB ajst-10994 7 10 , , PUNCT ajst-10994 7 11 which which PRON ajst-10994 7 12 incorporates incorporate VERB ajst-10994 7 13 modules module NOUN ajst-10994 7 14 like like ADP ajst-10994 7 15 time time NOUN ajst-10994 7 16 series series PROPN ajst-10994 7 17 embedding embed VERB ajst-10994 7 18 , , PUNCT ajst-10994 7 19 attention attention NOUN ajst-10994 7 20 mechanism mechanism NOUN ajst-10994 7 21 , , PUNCT ajst-10994 7 22 and and CCONJ ajst-10994 7 23 multilayer multilayer ADJ ajst-10994 7 24 perceptron perceptron PROPN ajst-10994 7 25 to to PART ajst-10994 7 26 promptly promptly ADV ajst-10994 7 27 approach approach VERB ajst-10994 7 28 1d 1d NUM ajst-10994 7 29 oscillatory oscillatory ADJ ajst-10994 7 30 motion motion NOUN ajst-10994 7 31 signal signal NOUN ajst-10994 7 32 data datum NOUN ajst-10994 7 33 . . PUNCT ajst-10994 8 1 subsequently subsequently ADV ajst-10994 8 2 , , PUNCT ajst-10994 8 3 the the DET ajst-10994 8 4 hypothetical hypothetical ADJ ajst-10994 8 5 model model NOUN ajst-10994 8 6 is be AUX ajst-10994 8 7 trained train VERB ajst-10994 8 8 and and CCONJ ajst-10994 8 9 tested test VERB ajst-10994 8 10 on on ADP ajst-10994 8 11 multiple multiple ADJ ajst-10994 8 12 public public ADJ ajst-10994 8 13 datasets dataset NOUN ajst-10994 8 14 , , PUNCT ajst-10994 8 15 and and CCONJ ajst-10994 8 16 its its PRON ajst-10994 8 17 fault fault NOUN ajst-10994 8 18 diagnosis diagnosis NOUN ajst-10994 8 19 results result NOUN ajst-10994 8 20 are be AUX ajst-10994 8 21 compared compare VERB ajst-10994 8 22 with with ADP ajst-10994 8 23 existing exist VERB ajst-10994 8 24 achievements achievement NOUN ajst-10994 8 25 in in ADP ajst-10994 8 26 the the DET ajst-10994 8 27 literature literature NOUN ajst-10994 8 28 . . PUNCT ajst-10994 9 1 the the DET ajst-10994 9 2 proficiency proficiency NOUN ajst-10994 9 3 of of ADP ajst-10994 9 4 the the DET ajst-10994 9 5 model model NOUN ajst-10994 9 6 is be AUX ajst-10994 9 7 thoroughly thoroughly ADV ajst-10994 9 8 verified verify VERB ajst-10994 9 9 , , PUNCT ajst-10994 9 10 and and CCONJ ajst-10994 9 11 the the DET ajst-10994 9 12 fault fault NOUN ajst-10994 9 13 diagnosis diagnosis NOUN ajst-10994 9 14 performance performance NOUN ajst-10994 9 15 of of ADP ajst-10994 9 16 this this DET ajst-10994 9 17 proposed propose VERB ajst-10994 9 18 approach approach NOUN ajst-10994 9 19 surpasses surpass VERB ajst-10994 9 20 many many ADJ ajst-10994 9 21 existing exist VERB ajst-10994 9 22 fault fault NOUN ajst-10994 9 23 diagnosis diagnosis NOUN ajst-10994 9 24 methods method NOUN ajst-10994 9 25 . . PUNCT ajst-10994 10 1 keywords keyword NOUN ajst-10994 10 2 : : PUNCT ajst-10994 10 3 rotating rotate VERB ajst-10994 10 4 machinery machinery NOUN ajst-10994 10 5 , , PUNCT ajst-10994 10 6 fault fault VERB ajst-10994 10 7 diagnosis diagnosis NOUN ajst-10994 10 8 method method NOUN ajst-10994 10 9 , , PUNCT ajst-10994 10 10 transformer transformer NOUN ajst-10994 10 11 , , PUNCT ajst-10994 10 12 attention attention NOUN ajst-10994 10 13 mechanism mechanism NOUN ajst-10994 10 14 . . PUNCT ajst-10994 11 1 1 1 X ajst-10994 11 2 . . X ajst-10994 11 3 introduction introduction NOUN ajst-10994 11 4 as as ADP ajst-10994 11 5 the the DET ajst-10994 11 6 advancement advancement NOUN ajst-10994 11 7 of of ADP ajst-10994 11 8 industry industry NOUN ajst-10994 11 9 and and CCONJ ajst-10994 11 10 technological technological ADJ ajst-10994 11 11 developments development NOUN ajst-10994 11 12 , , PUNCT ajst-10994 11 13 progressive progressive ADJ ajst-10994 11 14 industrial industrial ADJ ajst-10994 11 15 machinery machinery NOUN ajst-10994 11 16 has have AUX ajst-10994 11 17 become become VERB ajst-10994 11 18 increasingly increasingly ADV ajst-10994 11 19 functional functional ADJ ajst-10994 11 20 and and CCONJ ajst-10994 11 21 complex complex ADJ ajst-10994 11 22 . . PUNCT ajst-10994 12 1 among among ADP ajst-10994 12 2 them they PRON ajst-10994 12 3 , , PUNCT ajst-10994 12 4 rotating rotate VERB ajst-10994 12 5 machinery machinery NOUN ajst-10994 12 6 is be AUX ajst-10994 12 7 one one NUM ajst-10994 12 8 of of ADP ajst-10994 12 9 the the DET ajst-10994 12 10 most most ADV ajst-10994 12 11 crucial crucial ADJ ajst-10994 12 12 components component NOUN ajst-10994 12 13 in in ADP ajst-10994 12 14 various various ADJ ajst-10994 12 15 industrial industrial ADJ ajst-10994 12 16 fields field NOUN ajst-10994 12 17 such such ADJ ajst-10994 12 18 as as ADP ajst-10994 12 19 aviation aviation NOUN ajst-10994 12 20 , , PUNCT ajst-10994 12 21 aerospace aerospace NOUN ajst-10994 12 22 , , PUNCT ajst-10994 12 23 and and CCONJ ajst-10994 12 24 maritime maritime NOUN ajst-10994 12 25 industries[1 industries[1 NUM ajst-10994 12 26 ] ] PUNCT ajst-10994 12 27 . . PUNCT ajst-10994 13 1 however however ADV ajst-10994 13 2 , , PUNCT ajst-10994 13 3 due due ADP ajst-10994 13 4 to to ADP ajst-10994 13 5 harsh harsh ADJ ajst-10994 13 6 operating operating NOUN ajst-10994 13 7 conditions condition NOUN ajst-10994 13 8 such such ADJ ajst-10994 13 9 as as ADP ajst-10994 13 10 heavy heavy ADJ ajst-10994 13 11 loads load NOUN ajst-10994 13 12 , , PUNCT ajst-10994 13 13 high high ADJ ajst-10994 13 14 temperatures temperature NOUN ajst-10994 13 15 , , PUNCT ajst-10994 13 16 and and CCONJ ajst-10994 13 17 high high ADJ ajst-10994 13 18 pressures pressure NOUN ajst-10994 13 19 , , PUNCT ajst-10994 13 20 some some DET ajst-10994 13 21 critical critical ADJ ajst-10994 13 22 elements element NOUN ajst-10994 13 23 of of ADP ajst-10994 13 24 rotating rotate VERB ajst-10994 13 25 equipment equipment NOUN ajst-10994 13 26 , , PUNCT ajst-10994 13 27 some some DET ajst-10994 13 28 examples example NOUN ajst-10994 13 29 are be AUX ajst-10994 13 30 bearings bearing NOUN ajst-10994 13 31 and and CCONJ ajst-10994 13 32 gears gear NOUN ajst-10994 13 33 , , PUNCT ajst-10994 13 34 are be AUX ajst-10994 13 35 prone prone ADJ ajst-10994 13 36 to to ADP ajst-10994 13 37 failures failure NOUN ajst-10994 13 38 during during ADP ajst-10994 13 39 operation operation NOUN ajst-10994 13 40 . . PUNCT ajst-10994 14 1 these these DET ajst-10994 14 2 failures failure NOUN ajst-10994 14 3 can can AUX ajst-10994 14 4 have have VERB ajst-10994 14 5 a a DET ajst-10994 14 6 notable notable ADJ ajst-10994 14 7 effect effect NOUN ajst-10994 14 8 on on ADP ajst-10994 14 9 the the DET ajst-10994 14 10 normal normal ADJ ajst-10994 14 11 operation operation NOUN ajst-10994 14 12 of of ADP ajst-10994 14 13 the the DET ajst-10994 14 14 equipment equipment NOUN ajst-10994 14 15 , , PUNCT ajst-10994 14 16 giving give VERB ajst-10994 14 17 rise rise NOUN ajst-10994 14 18 to to ADP ajst-10994 14 19 a a DET ajst-10994 14 20 decline decline NOUN ajst-10994 14 21 in in ADP ajst-10994 14 22 service service NOUN ajst-10994 14 23 performance performance NOUN ajst-10994 14 24 and and CCONJ ajst-10994 14 25 , , PUNCT ajst-10994 14 26 in in ADP ajst-10994 14 27 extreme extreme ADJ ajst-10994 14 28 cases case NOUN ajst-10994 14 29 , , PUNCT ajst-10994 14 30 mechanical mechanical ADJ ajst-10994 14 31 failure failure NOUN ajst-10994 14 32 or or CCONJ ajst-10994 14 33 disintegration disintegration NOUN ajst-10994 14 34 . . PUNCT ajst-10994 15 1 this this PRON ajst-10994 15 2 poses pose VERB ajst-10994 15 3 serious serious ADJ ajst-10994 15 4 risks risk NOUN ajst-10994 15 5 to to ADP ajst-10994 15 6 manufacturing manufacture VERB ajst-10994 15 7 quality quality NOUN ajst-10994 15 8 and and CCONJ ajst-10994 15 9 operational operational ADJ ajst-10994 15 10 safety safety NOUN ajst-10994 15 11 . . PUNCT ajst-10994 16 1 therefore therefore ADV ajst-10994 16 2 , , PUNCT ajst-10994 16 3 fault fault VERB ajst-10994 16 4 diagnosis diagnosis NOUN ajst-10994 16 5 and and CCONJ ajst-10994 16 6 health health NOUN ajst-10994 16 7 watchfulness watchfulness NOUN ajst-10994 16 8 on on ADP ajst-10994 16 9 rotating rotate VERB ajst-10994 16 10 machinery machinery NOUN ajst-10994 16 11 are be AUX ajst-10994 16 12 essential essential ADJ ajst-10994 16 13 aspects aspect NOUN ajst-10994 16 14 in in ADP ajst-10994 16 15 industrial industrial ADJ ajst-10994 16 16 system system NOUN ajst-10994 16 17 design design NOUN ajst-10994 16 18 and and CCONJ ajst-10994 16 19 maintenance[2 maintenance[2 NOUN ajst-10994 16 20 ] ] X ajst-10994 16 21 . . PUNCT ajst-10994 17 1 the the DET ajst-10994 17 2 theory theory NOUN ajst-10994 17 3 and and CCONJ ajst-10994 17 4 methods method NOUN ajst-10994 17 5 related relate VERB ajst-10994 17 6 to to ADP ajst-10994 17 7 machine machine NOUN ajst-10994 17 8 learning learning NOUN ajst-10994 17 9 have have AUX ajst-10994 17 10 undergone undergo VERB ajst-10994 17 11 many many ADJ ajst-10994 17 12 years year NOUN ajst-10994 17 13 of of ADP ajst-10994 17 14 development development NOUN ajst-10994 17 15 and and CCONJ ajst-10994 17 16 have have AUX ajst-10994 17 17 been be AUX ajst-10994 17 18 extensively extensively ADV ajst-10994 17 19 adopted adopt VERB ajst-10994 17 20 in in ADP ajst-10994 17 21 various various ADJ ajst-10994 17 22 fields field NOUN ajst-10994 17 23 , , PUNCT ajst-10994 17 24 including include VERB ajst-10994 17 25 fault fault NOUN ajst-10994 17 26 diagnosis diagnosis NOUN ajst-10994 17 27 . . PUNCT ajst-10994 18 1 currently currently ADV ajst-10994 18 2 , , PUNCT ajst-10994 18 3 prevailing prevail VERB ajst-10994 18 4 machine machine NOUN ajst-10994 18 5 learning learning NOUN ajst-10994 18 6 applications application NOUN ajst-10994 18 7 used use VERB ajst-10994 18 8 in in ADP ajst-10994 18 9 the the DET ajst-10994 18 10 field field NOUN ajst-10994 18 11 of of ADP ajst-10994 18 12 fault fault NOUN ajst-10994 18 13 diagnosis diagnosis NOUN ajst-10994 18 14 include include VERB ajst-10994 18 15 svm svm PROPN ajst-10994 18 16 , , PUNCT ajst-10994 18 17 knn knn PROPN ajst-10994 18 18 , , PUNCT ajst-10994 18 19 som som NOUN ajst-10994 18 20 , , PUNCT ajst-10994 18 21 and and CCONJ ajst-10994 18 22 back back ADJ ajst-10994 18 23 propagation propagation NOUN ajst-10994 18 24 neural neural ADJ ajst-10994 18 25 network network NOUN ajst-10994 18 26 ( ( PUNCT ajst-10994 18 27 bpnn bpnn NOUN ajst-10994 18 28 ) ) PUNCT ajst-10994 18 29 consisting consist VERB ajst-10994 18 30 of of ADP ajst-10994 18 31 fully fully ADV ajst-10994 18 32 connected connect VERB ajst-10994 18 33 layers[3 layers[3 PROPN ajst-10994 18 34 ] ] PUNCT ajst-10994 18 35 . . PUNCT ajst-10994 19 1 yan yan PROPN ajst-10994 19 2 et et PROPN ajst-10994 19 3 al al PROPN ajst-10994 19 4 . . PROPN ajst-10994 19 5 utilized utilized PROPN ajst-10994 19 6 vmd vmd PROPN ajst-10994 19 7 to to PART ajst-10994 19 8 manipulate manipulate VERB ajst-10994 19 9 vibration vibration NOUN ajst-10994 19 10 signals signal NOUN ajst-10994 19 11 from from ADP ajst-10994 19 12 bearings bearing NOUN ajst-10994 19 13 , , PUNCT ajst-10994 19 14 combining combine VERB ajst-10994 19 15 time time NOUN ajst-10994 19 16 - - PUNCT ajst-10994 19 17 domain domain NOUN ajst-10994 19 18 and and CCONJ ajst-10994 19 19 frequency frequency NOUN ajst-10994 19 20 - - PUNCT ajst-10994 19 21 domain domain NOUN ajst-10994 19 22 features feature NOUN ajst-10994 19 23 to to PART ajst-10994 19 24 extract extract VERB ajst-10994 19 25 multi multi ADJ ajst-10994 19 26 - - NOUN ajst-10994 19 27 attribute attribute NOUN ajst-10994 19 28 , , PUNCT ajst-10994 19 29 timevarying timevarye VERB ajst-10994 19 30 spectrum spectrum NOUN ajst-10994 19 31 domain domain NOUN ajst-10994 19 32 properties property NOUN ajst-10994 19 33 . . PUNCT ajst-10994 20 1 subsequently subsequently ADV ajst-10994 20 2 , , PUNCT ajst-10994 20 3 they they PRON ajst-10994 20 4 employed employ VERB ajst-10994 20 5 the the DET ajst-10994 20 6 laplacian laplacian ADJ ajst-10994 20 7 score score NOUN ajst-10994 20 8 ( ( PUNCT ajst-10994 20 9 ls ls PROPN ajst-10994 20 10 ) ) PUNCT ajst-10994 20 11 to to PART ajst-10994 20 12 select select VERB ajst-10994 20 13 and and CCONJ ajst-10994 20 14 reduce reduce VERB ajst-10994 20 15 the the DET ajst-10994 20 16 dimensionality dimensionality NOUN ajst-10994 20 17 of of ADP ajst-10994 20 18 the the DET ajst-10994 20 19 extracted extract VERB ajst-10994 20 20 high high ADJ ajst-10994 20 21 - - PUNCT ajst-10994 20 22 dimensional dimensional ADJ ajst-10994 20 23 timefrequency timefrequency NOUN ajst-10994 20 24 domain domain NOUN ajst-10994 20 25 features feature NOUN ajst-10994 20 26 . . PUNCT ajst-10994 21 1 the the DET ajst-10994 21 2 reduced reduce VERB ajst-10994 21 3 features feature NOUN ajst-10994 21 4 were be AUX ajst-10994 21 5 used use VERB ajst-10994 21 6 as as ADP ajst-10994 21 7 inputs input NOUN ajst-10994 21 8 for for ADP ajst-10994 21 9 support support NOUN ajst-10994 21 10 vector vector NOUN ajst-10994 21 11 machine machine NOUN ajst-10994 21 12 ( ( PUNCT ajst-10994 21 13 svm svm PROPN ajst-10994 21 14 ) ) PUNCT ajst-10994 21 15 , , PUNCT ajst-10994 21 16 and and CCONJ ajst-10994 21 17 the the DET ajst-10994 21 18 svm svm ADJ ajst-10994 21 19 criteria criterion NOUN ajst-10994 21 20 were be AUX ajst-10994 21 21 refined refine VERB ajst-10994 21 22 using use VERB ajst-10994 21 23 pso pso NOUN ajst-10994 21 24 . . PUNCT ajst-10994 22 1 this this DET ajst-10994 22 2 approach approach NOUN ajst-10994 22 3 achieved achieve VERB ajst-10994 22 4 effective effective ADJ ajst-10994 22 5 fault fault NOUN ajst-10994 22 6 diagnosis diagnosis NOUN ajst-10994 22 7 of of ADP ajst-10994 22 8 bearings bearing NOUN ajst-10994 22 9 and and CCONJ ajst-10994 22 10 demonstrated demonstrate VERB ajst-10994 22 11 favorable favorable ADJ ajst-10994 22 12 fault fault NOUN ajst-10994 22 13 diagnostic diagnostic ADJ ajst-10994 22 14 performance performance NOUN ajst-10994 22 15 . . PUNCT ajst-10994 23 1 janssens janssen NOUN ajst-10994 23 2 et et PROPN ajst-10994 23 3 al al PROPN ajst-10994 23 4 . . PROPN ajst-10994 23 5 proposed propose VERB ajst-10994 23 6 a a DET ajst-10994 23 7 bearing bear VERB ajst-10994 23 8 fault fault NOUN ajst-10994 23 9 diagnosis diagnosis NOUN ajst-10994 23 10 tactics tactic NOUN ajst-10994 23 11 that that PRON ajst-10994 23 12 combines combine VERB ajst-10994 23 13 the the DET ajst-10994 23 14 fcm fcm NOUN ajst-10994 23 15 with with ADP ajst-10994 23 16 k k X ajst-10994 23 17 - - PUNCT ajst-10994 23 18 nearest near ADJ ajst-10994 23 19 neighbor neighbor NOUN ajst-10994 23 20 ( ( PUNCT ajst-10994 23 21 knn knn PROPN ajst-10994 23 22 ) ) PUNCT ajst-10994 23 23 . . PUNCT ajst-10994 24 1 in in ADP ajst-10994 24 2 this this DET ajst-10994 24 3 method method NOUN ajst-10994 24 4 , , PUNCT ajst-10994 24 5 pso pso NOUN ajst-10994 24 6 ( ( PUNCT ajst-10994 24 7 particle particle NOUN ajst-10994 24 8 swarm swarm NOUN ajst-10994 24 9 optimization optimization NOUN ajst-10994 24 10 ) ) PUNCT ajst-10994 24 11 was be AUX ajst-10994 24 12 also also ADV ajst-10994 24 13 utilized utilize VERB ajst-10994 24 14 to to PART ajst-10994 24 15 optimize optimize VERB ajst-10994 24 16 the the DET ajst-10994 24 17 knn knn PROPN ajst-10994 24 18 algorithm algorithm PROPN ajst-10994 24 19 , , PUNCT ajst-10994 24 20 reducing reduce VERB ajst-10994 24 21 computational computational ADJ ajst-10994 24 22 complexity[4 complexity[4 PROPN ajst-10994 24 23 ] ] PUNCT ajst-10994 24 24 . . PUNCT ajst-10994 25 1 comparative comparative ADJ ajst-10994 25 2 test test NOUN ajst-10994 25 3 findings finding NOUN ajst-10994 25 4 demonstrated demonstrate VERB ajst-10994 25 5 that that SCONJ ajst-10994 25 6 this this DET ajst-10994 25 7 approach approach NOUN ajst-10994 25 8 can can AUX ajst-10994 25 9 properly properly ADV ajst-10994 25 10 classify classify VERB ajst-10994 25 11 the the DET ajst-10994 25 12 major major ADJ ajst-10994 25 13 fault fault NOUN ajst-10994 25 14 groupings grouping NOUN ajst-10994 25 15 of of ADP ajst-10994 25 16 bearings bearing NOUN ajst-10994 25 17 using use VERB ajst-10994 25 18 a a DET ajst-10994 25 19 limited limited ADJ ajst-10994 25 20 amount amount NOUN ajst-10994 25 21 of of ADP ajst-10994 25 22 fault fault NOUN ajst-10994 25 23 data datum NOUN ajst-10994 25 24 . . PUNCT ajst-10994 26 1 zhang zhang PROPN ajst-10994 26 2 et et PROPN ajst-10994 26 3 al al PROPN ajst-10994 26 4 . . PROPN ajst-10994 26 5 utilized utilize VERB ajst-10994 26 6 the the DET ajst-10994 26 7 chaotic chaotic ADJ ajst-10994 26 8 adaptive adaptive ADJ ajst-10994 26 9 gravitational gravitational ADJ ajst-10994 26 10 search search NOUN ajst-10994 26 11 algorithm algorithm NOUN ajst-10994 26 12 and and CCONJ ajst-10994 26 13 pso pso NOUN ajst-10994 26 14 to to PART ajst-10994 26 15 optimize optimize VERB ajst-10994 26 16 the the DET ajst-10994 26 17 training training NOUN ajst-10994 26 18 process process NOUN ajst-10994 26 19 of of ADP ajst-10994 26 20 the the DET ajst-10994 26 21 bpnn bpnn NOUN ajst-10994 26 22 . . PUNCT ajst-10994 27 1 they they PRON ajst-10994 27 2 operationalized operationalize VERB ajst-10994 27 3 this this DET ajst-10994 27 4 approach approach NOUN ajst-10994 27 5 to to ADP ajst-10994 27 6 the the DET ajst-10994 27 7 fault fault NOUN ajst-10994 27 8 diagnosis diagnosis NOUN ajst-10994 27 9 problem problem NOUN ajst-10994 27 10 of of ADP ajst-10994 27 11 motor motor NOUN ajst-10994 27 12 drive drive NOUN ajst-10994 27 13 systems system NOUN ajst-10994 27 14 and and CCONJ ajst-10994 27 15 found find VERB ajst-10994 27 16 that that SCONJ ajst-10994 27 17 introducing introduce VERB ajst-10994 27 18 adaptive adaptive ADJ ajst-10994 27 19 gravitational gravitational ADJ ajst-10994 27 20 constant constant ADJ ajst-10994 27 21 decay decay NOUN ajst-10994 27 22 and and CCONJ ajst-10994 27 23 chaotic chaotic ADJ ajst-10994 27 24 mapping mapping NOUN ajst-10994 27 25 during during ADP ajst-10994 27 26 the the DET ajst-10994 27 27 training training NOUN ajst-10994 27 28 process process NOUN ajst-10994 27 29 can can AUX ajst-10994 27 30 enhance enhance VERB ajst-10994 27 31 the the DET ajst-10994 27 32 pattern pattern NOUN ajst-10994 27 33 recognition recognition NOUN ajst-10994 27 34 performance performance NOUN ajst-10994 27 35 of of ADP ajst-10994 27 36 bpnn[5 bpnn[5 NOUN ajst-10994 27 37 ] ] PUNCT ajst-10994 27 38 . . PUNCT ajst-10994 28 1 this this DET ajst-10994 28 2 paper paper NOUN ajst-10994 28 3 presents present VERB ajst-10994 28 4 an an DET ajst-10994 28 5 improved improved ADJ ajst-10994 28 6 temporal temporal ADJ ajst-10994 28 7 dataset dataset NOUN ajst-10994 28 8 transformer transformer NOUN ajst-10994 28 9 - - PUNCT ajst-10994 28 10 based base VERB ajst-10994 28 11 fault fault NOUN ajst-10994 28 12 diagnosis diagnosis NOUN ajst-10994 28 13 method method NOUN ajst-10994 28 14 for for ADP ajst-10994 28 15 rotating rotate VERB ajst-10994 28 16 machinery machinery NOUN ajst-10994 28 17 . . PUNCT ajst-10994 29 1 firstly firstly ADV ajst-10994 29 2 , , PUNCT ajst-10994 29 3 a a DET ajst-10994 29 4 temporal temporal ADJ ajst-10994 29 5 dataset dataset ADJ ajst-10994 29 6 tokenizer tokenizer NOUN ajst-10994 29 7 is be AUX ajst-10994 29 8 designed design VERB ajst-10994 29 9 to to PART ajst-10994 29 10 directly directly ADV ajst-10994 29 11 process process VERB ajst-10994 29 12 1d 1d NUM ajst-10994 29 13 format format NOUN ajst-10994 29 14 vibration vibration NOUN ajst-10994 29 15 signal signal NOUN ajst-10994 29 16 data datum NOUN ajst-10994 29 17 , , PUNCT ajst-10994 29 18 and and CCONJ ajst-10994 29 19 a a DET ajst-10994 29 20 multihead multihead NOUN ajst-10994 29 21 self self NOUN ajst-10994 29 22 - - PUNCT ajst-10994 29 23 attention attention NOUN ajst-10994 29 24 mechanism mechanism NOUN ajst-10994 29 25 is be AUX ajst-10994 29 26 utilized utilize VERB ajst-10994 29 27 to to PART ajst-10994 29 28 establish establish VERB ajst-10994 29 29 the the DET ajst-10994 29 30 time time NOUN ajst-10994 29 31 series series PROPN ajst-10994 29 32 transformer transformer PROPN ajst-10994 29 33 model model NOUN ajst-10994 29 34 . . PUNCT ajst-10994 30 1 subsequently subsequently ADV ajst-10994 30 2 , , PUNCT ajst-10994 30 3 the the DET ajst-10994 30 4 model model NOUN ajst-10994 30 5 is be AUX ajst-10994 30 6 trained train VERB ajst-10994 30 7 and and CCONJ ajst-10994 30 8 tested test VERB ajst-10994 30 9 on on ADP ajst-10994 30 10 publicly publicly ADV ajst-10994 30 11 reachable reachable ADJ ajst-10994 30 12 experimental experimental ADJ ajst-10994 30 13 datasets dataset NOUN ajst-10994 30 14 , , PUNCT ajst-10994 30 15 and and CCONJ ajst-10994 30 16 the the DET ajst-10994 30 17 testing testing NOUN ajst-10994 30 18 results result VERB ajst-10994 30 19 across across ADP ajst-10994 30 20 multiple multiple ADJ ajst-10994 30 21 datasets dataset NOUN ajst-10994 30 22 are be AUX ajst-10994 30 23 provided provide VERB ajst-10994 30 24 to to PART ajst-10994 30 25 analyze analyze VERB ajst-10994 30 26 and and CCONJ ajst-10994 30 27 test test VERB ajst-10994 30 28 the the DET ajst-10994 30 29 performance performance NOUN ajst-10994 30 30 of of ADP ajst-10994 30 31 the the DET ajst-10994 30 32 suggested suggest VERB ajst-10994 30 33 methodology methodology NOUN ajst-10994 30 34 . . PUNCT ajst-10994 31 1 through through ADP ajst-10994 31 2 this this DET ajst-10994 31 3 validation validation NOUN ajst-10994 31 4 , , PUNCT ajst-10994 31 5 fault fault VERB ajst-10994 31 6 diagnosis diagnosis NOUN ajst-10994 31 7 performance performance NOUN ajst-10994 31 8 of of ADP ajst-10994 31 9 this this DET ajst-10994 31 10 approach approach NOUN ajst-10994 31 11 outperforms outperform VERB ajst-10994 31 12 many many ADJ ajst-10994 31 13 existing exist VERB ajst-10994 31 14 fault fault NOUN ajst-10994 31 15 diagnosis diagnosis NOUN ajst-10994 31 16 methods method NOUN ajst-10994 31 17 . . PUNCT ajst-10994 32 1 2 2 NUM ajst-10994 32 2 . . NOUN ajst-10994 32 3 time time NOUN ajst-10994 32 4 series series PROPN ajst-10994 32 5 transformer transformer PROPN ajst-10994 32 6 model model NOUN ajst-10994 32 7 based base VERB ajst-10994 32 8 on on ADP ajst-10994 32 9 mechanical mechanical ADJ ajst-10994 32 10 fault fault NOUN ajst-10994 32 11 diagnosis diagnosis NOUN ajst-10994 32 12 in in ADP ajst-10994 32 13 experiments experiment NOUN ajst-10994 32 14 , , PUNCT ajst-10994 32 15 the the DET ajst-10994 32 16 collected collect VERB ajst-10994 32 17 raw raw ADJ ajst-10994 32 18 vibration vibration NOUN ajst-10994 32 19 signals signal NOUN ajst-10994 32 20 are be AUX ajst-10994 32 21 typically typically ADV ajst-10994 32 22 1d 1d NUM ajst-10994 32 23 format format NOUN ajst-10994 32 24 time time NOUN ajst-10994 32 25 series series PROPN ajst-10994 32 26 data data PROPN ajst-10994 32 27 . . PUNCT ajst-10994 33 1 however however ADV ajst-10994 33 2 , , PUNCT ajst-10994 33 3 the the DET ajst-10994 33 4 native native ADJ ajst-10994 33 5 transformer transformer NOUN ajst-10994 33 6 models model NOUN ajst-10994 33 7 widely widely ADV ajst-10994 33 8 used use VERB ajst-10994 33 9 in in ADP ajst-10994 33 10 nlp nlp NOUN ajst-10994 33 11 ( ( PUNCT ajst-10994 33 12 natural natural ADJ ajst-10994 33 13 language language NOUN ajst-10994 33 14 processing processing NOUN ajst-10994 33 15 ) ) PUNCT ajst-10994 33 16 domain domain NOUN ajst-10994 33 17 take take VERB ajst-10994 33 18 tokenized tokenized ADJ ajst-10994 33 19 text text NOUN ajst-10994 33 20 sequences sequence NOUN ajst-10994 33 21 as as ADP ajst-10994 33 22 input input NOUN ajst-10994 33 23 , , PUNCT ajst-10994 33 24 and and CCONJ ajst-10994 33 25 the the DET ajst-10994 33 26 vision vision NOUN ajst-10994 33 27 transformer transformer NOUN ajst-10994 33 28 ( ( PUNCT ajst-10994 33 29 vit vit NOUN ajst-10994 33 30 ) ) PUNCT ajst-10994 33 31 used use VERB ajst-10994 33 32 in in ADP ajst-10994 33 33 the the DET ajst-10994 33 34 computer computer NOUN ajst-10994 33 35 vision vision NOUN ajst-10994 33 36 domain domain NOUN ajst-10994 33 37 processes process VERB ajst-10994 33 38 2d 2d PROPN ajst-10994 33 39 format format NOUN ajst-10994 33 40 rgb rgb PROPN ajst-10994 33 41 images image NOUN ajst-10994 33 42 with with ADP ajst-10994 33 43 three three NUM ajst-10994 33 44 channels channel NOUN ajst-10994 33 45 . . PUNCT ajst-10994 34 1 these these DET ajst-10994 34 2 transformer transformer NOUN ajst-10994 34 3 models model NOUN ajst-10994 34 4 can can AUX ajst-10994 34 5 not not PART ajst-10994 34 6 directly directly ADV ajst-10994 34 7 handle handle VERB ajst-10994 34 8 1d 1d NUM ajst-10994 34 9 format format NOUN ajst-10994 34 10 80 80 NUM ajst-10994 34 11 vibration vibration NOUN ajst-10994 34 12 signal signal NOUN ajst-10994 34 13 data datum NOUN ajst-10994 34 14 . . PUNCT ajst-10994 35 1 within within ADP ajst-10994 35 2 this this DET ajst-10994 35 3 chapter chapter NOUN ajst-10994 35 4 's 's PART ajst-10994 35 5 scope scope NOUN ajst-10994 35 6 , , PUNCT ajst-10994 35 7 we we PRON ajst-10994 35 8 propose propose VERB ajst-10994 35 9 a a DET ajst-10994 35 10 transformer transformer NOUN ajst-10994 35 11 model model NOUN ajst-10994 35 12 unambiguously unambiguously ADV ajst-10994 35 13 constructed construct VERB ajst-10994 35 14 to to PART ajst-10994 35 15 process process VERB ajst-10994 35 16 raw raw ADJ ajst-10994 35 17 vibration vibration NOUN ajst-10994 35 18 signal signal NOUN ajst-10994 35 19 data datum NOUN ajst-10994 35 20 , , PUNCT ajst-10994 35 21 called call VERB ajst-10994 35 22 time time NOUN ajst-10994 35 23 series series PROPN ajst-10994 35 24 transformer transformer NOUN ajst-10994 35 25 ( ( PUNCT ajst-10994 35 26 tst tst NOUN ajst-10994 35 27 ) ) PUNCT ajst-10994 35 28 . . PUNCT ajst-10994 36 1 as as SCONJ ajst-10994 36 2 shown show VERB ajst-10994 36 3 in in ADP ajst-10994 36 4 figure figure NOUN ajst-10994 36 5 1 1 NUM ajst-10994 36 6 , , PUNCT ajst-10994 36 7 the the DET ajst-10994 36 8 tst tst NOUN ajst-10994 36 9 model model NOUN ajst-10994 36 10 comprises comprise VERB ajst-10994 36 11 components component NOUN ajst-10994 36 12 such such ADJ ajst-10994 36 13 as as ADP ajst-10994 36 14 time time NOUN ajst-10994 36 15 series series PROPN ajst-10994 36 16 tokenization tokenization NOUN ajst-10994 36 17 , , PUNCT ajst-10994 36 18 transformer transformer NOUN ajst-10994 36 19 layers layer NOUN ajst-10994 36 20 , , PUNCT ajst-10994 36 21 and and CCONJ ajst-10994 36 22 classifier classifier NOUN ajst-10994 36 23 layers[6 layers[6 PROPN ajst-10994 36 24 ] ] PUNCT ajst-10994 36 25 . . PUNCT ajst-10994 37 1 figure figure NOUN ajst-10994 37 2 1 1 NUM ajst-10994 37 3 . . PUNCT ajst-10994 38 1 the the DET ajst-10994 38 2 composition composition NOUN ajst-10994 38 3 of of ADP ajst-10994 38 4 the the DET ajst-10994 38 5 planed plane VERB ajst-10994 38 6 tst tst NOUN ajst-10994 38 7 paragon paragon NOUN ajst-10994 38 8 2.1 2.1 NUM ajst-10994 38 9 . . PUNCT ajst-10994 38 10 time time PROPN ajst-10994 38 11 series series PROPN ajst-10994 38 12 segmentation segmentation PROPN ajst-10994 38 13 2.1.1 2.1.1 NUM ajst-10994 38 14 . . PUNCT ajst-10994 38 15 time time NOUN ajst-10994 38 16 series series PROPN ajst-10994 38 17 embedding embed VERB ajst-10994 38 18 in in ADP ajst-10994 38 19 general general ADJ ajst-10994 38 20 , , PUNCT ajst-10994 38 21 a a DET ajst-10994 38 22 batch batch NOUN ajst-10994 38 23 of of ADP ajst-10994 38 24 time time NOUN ajst-10994 38 25 - - PUNCT ajst-10994 38 26 ordered order VERB ajst-10994 38 27 sequence sequence NOUN ajst-10994 38 28 can can AUX ajst-10994 38 29 be be AUX ajst-10994 38 30 represented represent VERB ajst-10994 38 31 as as ADP ajst-10994 38 32 ๐‘ก ๐‘ก PROPN ajst-10994 38 33 โˆˆ โˆˆ PROPN ajst-10994 38 34 rbร—l rbร—l ADV ajst-10994 38 35 , , PUNCT ajst-10994 38 36 where where SCONJ ajst-10994 38 37 b b NOUN ajst-10994 38 38 is be AUX ajst-10994 38 39 the the DET ajst-10994 38 40 batch batch NOUN ajst-10994 38 41 capacity capacity NOUN ajst-10994 38 42 , , PUNCT ajst-10994 38 43 and and CCONJ ajst-10994 38 44 l l NOUN ajst-10994 38 45 represents represent VERB ajst-10994 38 46 the the DET ajst-10994 38 47 examined examine VERB ajst-10994 38 48 time time NOUN ajst-10994 38 49 series series PROPN ajst-10994 38 50 ' ' PART ajst-10994 38 51 sampling sample VERB ajst-10994 38 52 length length NOUN ajst-10994 38 53 . . PUNCT ajst-10994 39 1 in in ADP ajst-10994 39 2 the the DET ajst-10994 39 3 process process NOUN ajst-10994 39 4 of of ADP ajst-10994 39 5 time time NOUN ajst-10994 39 6 series series PROPN ajst-10994 39 7 transformation transformation NOUN ajst-10994 39 8 , , PUNCT ajst-10994 39 9 initially initially ADV ajst-10994 39 10 , , PUNCT ajst-10994 39 11 related relate VERB ajst-10994 39 12 to to ADP ajst-10994 39 13 the the DET ajst-10994 39 14 tokenization tokenization NOUN ajst-10994 39 15 process process NOUN ajst-10994 39 16 in in ADP ajst-10994 39 17 nlp nlp NOUN ajst-10994 39 18 , , PUNCT ajst-10994 39 19 the the DET ajst-10994 39 20 received receive VERB ajst-10994 39 21 time time NOUN ajst-10994 39 22 - - PUNCT ajst-10994 39 23 ordered order VERB ajst-10994 39 24 sequence sequence NOUN ajst-10994 39 25 is be AUX ajst-10994 39 26 divided divide VERB ajst-10994 39 27 into into ADP ajst-10994 39 28 sub sub NOUN ajst-10994 39 29 - - NOUN ajst-10994 39 30 sequences sequence NOUN ajst-10994 39 31 of of ADP ajst-10994 39 32 specified specified ADJ ajst-10994 39 33 lengths length NOUN ajst-10994 39 34 . . PUNCT ajst-10994 40 1 these these DET ajst-10994 40 2 sub sub NOUN ajst-10994 40 3 - - NOUN ajst-10994 40 4 sequences sequence NOUN ajst-10994 40 5 are be AUX ajst-10994 40 6 then then ADV ajst-10994 40 7 concatenated concatenate VERB ajst-10994 40 8 in in ADP ajst-10994 40 9 order order NOUN ajst-10994 40 10 to to PART ajst-10994 40 11 form form VERB ajst-10994 40 12 a a DET ajst-10994 40 13 three three NUM ajst-10994 40 14 - - PUNCT ajst-10994 40 15 dimensional dimensional ADJ ajst-10994 40 16 tensor tensor NOUN ajst-10994 40 17 , , PUNCT ajst-10994 40 18 denoted denote VERB ajst-10994 40 19 as as ADP ajst-10994 40 20 [ [ X ajst-10994 40 21 t1 t1 PROPN ajst-10994 40 22 s s PROPN ajst-10994 40 23 , , PUNCT ajst-10994 40 24 t2 t2 PROPN ajst-10994 40 25 s s PROPN ajst-10994 40 26 , , PUNCT ajst-10994 40 27 t3 t3 PROPN ajst-10994 40 28 s s PROPN ajst-10994 40 29 , , PUNCT ajst-10994 40 30 โ€ฆ โ€ฆ PUNCT ajst-10994 40 31 , , PUNCT ajst-10994 40 32 tn tn PROPN ajst-10994 40 33 s s PART ajst-10994 40 34 ] ] X ajst-10994 40 35 โˆˆ โˆˆ PROPN ajst-10994 40 36 r๐ตร—๐‘๐‘  r๐ตร—๐‘๐‘  PROPN ajst-10994 40 37 ร—(๐ฟ/๐‘๐‘  ร—(๐ฟ/๐‘๐‘  PROPN ajst-10994 40 38 ) ) PUNCT ajst-10994 40 39 , , PUNCT ajst-10994 40 40 where where SCONJ ajst-10994 40 41 ns ns PROPN ajst-10994 40 42 represents represent VERB ajst-10994 40 43 the the DET ajst-10994 40 44 number number NOUN ajst-10994 40 45 of of ADP ajst-10994 40 46 segmented segment VERB ajst-10994 40 47 subsequences subsequence NOUN ajst-10994 40 48 , , PUNCT ajst-10994 40 49 and and CCONJ ajst-10994 40 50 l l NOUN ajst-10994 40 51 / / SYM ajst-10994 40 52 ns ns PROPN ajst-10994 40 53 should should AUX ajst-10994 40 54 be be AUX ajst-10994 40 55 an an DET ajst-10994 40 56 integer integer NOUN ajst-10994 40 57 . . PUNCT ajst-10994 41 1 next next ADV ajst-10994 41 2 , , PUNCT ajst-10994 41 3 akin akin ADJ ajst-10994 41 4 to to ADP ajst-10994 41 5 the the DET ajst-10994 41 6 word word NOUN ajst-10994 41 7 embedding embed VERB ajst-10994 41 8 process process NOUN ajst-10994 41 9 , , PUNCT ajst-10994 41 10 each each DET ajst-10994 41 11 sub sub NOUN ajst-10994 41 12 - - NOUN ajst-10994 41 13 sequence sequence NOUN ajst-10994 41 14 is be AUX ajst-10994 41 15 mapped map VERB ajst-10994 41 16 to to ADP ajst-10994 41 17 a a DET ajst-10994 41 18 high high ADJ ajst-10994 41 19 - - PUNCT ajst-10994 41 20 dimensional dimensional ADJ ajst-10994 41 21 transformation transformation NOUN ajst-10994 41 22 space space NOUN ajst-10994 41 23 through through ADP ajst-10994 41 24 a a DET ajst-10994 41 25 simple simple ADJ ajst-10994 41 26 rectilinear rectilinear ADJ ajst-10994 41 27 transformation transformation NOUN ajst-10994 41 28 , , PUNCT ajst-10994 41 29 as as SCONJ ajst-10994 41 30 shown show VERB ajst-10994 41 31 in in ADP ajst-10994 41 32 equation equation NOUN ajst-10994 41 33 ( ( PUNCT ajst-10994 41 34 1 1 NUM ajst-10994 41 35 ) ) PUNCT ajst-10994 41 36 . . PUNCT ajst-10994 42 1 1 1 NUM ajst-10994 42 2 2 2 NUM ajst-10994 42 3 3 3 NUM ajst-10994 42 4 dim dim NOUN ajst-10994 42 5 [ [ PUNCT ajst-10994 42 6 , , PUNCT ajst-10994 42 7 , , PUNCT ajst-10994 42 8 , , PUNCT ajst-10994 42 9 ... ... PUNCT ajst-10994 42 10 , , PUNCT ajst-10994 42 11 ] ] X ajst-10994 42 12 s s X ajst-10994 42 13 s s X ajst-10994 42 14 n n NOUN ajst-10994 42 15 s s NOUN ajst-10994 42 16 s s X ajst-10994 42 17 s s X ajst-10994 42 18 s s NOUN ajst-10994 42 19 embedding embed VERB ajst-10994 42 20 b b PROPN ajst-10994 42 21 n n PRON ajst-10994 42 22 tokensseq tokensseq NOUN ajst-10994 42 23 t t PROPN ajst-10994 42 24 t t PROPN ajst-10994 42 25 t t PROPN ajst-10994 42 26 t t PROPN ajst-10994 42 27 w w NOUN ajst-10994 42 28 r r NOUN ajst-10994 42 29 ๏‚ด ๏‚ด NOUN ajst-10994 42 30 ๏‚ด ๏‚ด PROPN ajst-10994 42 31 ๏€ฝ ๏€ฝ PROPN ajst-10994 42 32 ๏‚ท ๏‚ท PRON ajst-10994 42 33 ๏ƒŽ ๏ƒŽ NOUN ajst-10994 42 34 ( ( PUNCT ajst-10994 42 35 1 1 NUM ajst-10994 42 36 ) ) PUNCT ajst-10994 42 37 where where SCONJ ajst-10994 42 38 wembedding wembedde VERB ajst-10994 42 39 โˆˆ โˆˆ PROPN ajst-10994 42 40 r r NOUN ajst-10994 42 41 represents represent VERB ajst-10994 42 42 the the DET ajst-10994 42 43 linear linear ADJ ajst-10994 42 44 mapping mapping NOUN ajst-10994 42 45 matrix matrix NOUN ajst-10994 42 46 , , PUNCT ajst-10994 42 47 which which PRON ajst-10994 42 48 is be AUX ajst-10994 42 49 a a DET ajst-10994 42 50 attainable attainable ADJ ajst-10994 42 51 parameter parameter NOUN ajst-10994 42 52 , , PUNCT ajst-10994 42 53 and and CCONJ ajst-10994 42 54 dim dim ADJ ajst-10994 42 55 denotes denote NOUN ajst-10994 42 56 the the DET ajst-10994 42 57 embedding embed VERB ajst-10994 42 58 time time NOUN ajst-10994 42 59 series series PROPN ajst-10994 42 60 measurement measurement PROPN ajst-10994 42 61 . . PUNCT ajst-10994 43 1 it it PRON ajst-10994 43 2 should should AUX ajst-10994 43 3 be be AUX ajst-10994 43 4 noted note VERB ajst-10994 43 5 that that SCONJ ajst-10994 43 6 , , PUNCT ajst-10994 43 7 in in ADP ajst-10994 43 8 order order NOUN ajst-10994 43 9 to to PART ajst-10994 43 10 enable enable VERB ajst-10994 43 11 the the DET ajst-10994 43 12 proposed propose VERB ajst-10994 43 13 tst tst NOUN ajst-10994 43 14 ( ( PUNCT ajst-10994 43 15 transformer transformer NOUN ajst-10994 43 16 - - PUNCT ajst-10994 43 17 based base VERB ajst-10994 43 18 time time NOUN ajst-10994 43 19 series series PROPN ajst-10994 43 20 ) ) PUNCT ajst-10994 43 21 model model NOUN ajst-10994 43 22 to to PART ajst-10994 43 23 learn learn VERB ajst-10994 43 24 more more ADJ ajst-10994 43 25 general general ADJ ajst-10994 43 26 mappings mapping NOUN ajst-10994 43 27 and and CCONJ ajst-10994 43 28 , , PUNCT ajst-10994 43 29 concurrent concurrent ADJ ajst-10994 43 30 with with ADV ajst-10994 43 31 , , PUNCT ajst-10994 43 32 reduce reduce VERB ajst-10994 43 33 the the DET ajst-10994 43 34 number number NOUN ajst-10994 43 35 of of ADP ajst-10994 43 36 model model NOUN ajst-10994 43 37 conditions condition NOUN ajst-10994 43 38 , , PUNCT ajst-10994 43 39 the the DET ajst-10994 43 40 tst tst NOUN ajst-10994 43 41 model model NOUN ajst-10994 43 42 employs employ VERB ajst-10994 43 43 unvarying unvarye VERB ajst-10994 43 44 linear linear ADJ ajst-10994 43 45 transformation transformation NOUN ajst-10994 43 46 matrix matrix NOUN ajst-10994 43 47 for for ADP ajst-10994 43 48 all all DET ajst-10994 43 49 subsequences subsequence NOUN ajst-10994 43 50 to to PART ajst-10994 43 51 perform perform VERB ajst-10994 43 52 the the DET ajst-10994 43 53 linear linear ADJ ajst-10994 43 54 transformation transformation NOUN ajst-10994 43 55 . . PUNCT ajst-10994 44 1 2.1.2 2.1.2 X ajst-10994 44 2 . . PUNCT ajst-10994 44 3 class class NOUN ajst-10994 44 4 tag tag NOUN ajst-10994 44 5 after after ADP ajst-10994 44 6 feature feature NOUN ajst-10994 44 7 extraction extraction NOUN ajst-10994 44 8 of of ADP ajst-10994 44 9 marker marker NOUN ajst-10994 44 10 sequences sequence NOUN ajst-10994 44 11 obtained obtain VERB ajst-10994 44 12 by by ADP ajst-10994 44 13 time time NOUN ajst-10994 44 14 series series PROPN ajst-10994 44 15 embedding embed VERB ajst-10994 44 16 based base VERB ajst-10994 44 17 on on ADP ajst-10994 44 18 attention attention NOUN ajst-10994 44 19 mechanism mechanism NOUN ajst-10994 44 20 , , PUNCT ajst-10994 44 21 the the DET ajst-10994 44 22 tst tst NOUN ajst-10994 44 23 model model NOUN ajst-10994 44 24 also also ADV ajst-10994 44 25 needs need VERB ajst-10994 44 26 to to PART ajst-10994 44 27 represent represent VERB ajst-10994 44 28 the the DET ajst-10994 44 29 extracted extract VERB ajst-10994 44 30 features feature NOUN ajst-10994 44 31 and and CCONJ ajst-10994 44 32 convert convert VERB ajst-10994 44 33 them they PRON ajst-10994 44 34 into into ADP ajst-10994 44 35 feature feature NOUN ajst-10994 44 36 maps map NOUN ajst-10994 44 37 . . PUNCT ajst-10994 45 1 at at ADP ajst-10994 45 2 present present ADJ ajst-10994 45 3 , , PUNCT ajst-10994 45 4 there there PRON ajst-10994 45 5 are be VERB ajst-10994 45 6 two two NUM ajst-10994 45 7 main main ADJ ajst-10994 45 8 methods method NOUN ajst-10994 45 9 to to PART ajst-10994 45 10 represent represent VERB ajst-10994 45 11 the the DET ajst-10994 45 12 features feature NOUN ajst-10994 45 13 of of ADP ajst-10994 45 14 the the DET ajst-10994 45 15 tag tag NOUN ajst-10994 45 16 sequence sequence NOUN ajst-10994 45 17 : : PUNCT ajst-10994 45 18 ( ( PUNCT ajst-10994 45 19 1 1 X ajst-10994 45 20 ) ) PUNCT ajst-10994 45 21 global global ADJ ajst-10994 45 22 pooling pooling NOUN ajst-10994 45 23 of of ADP ajst-10994 45 24 the the DET ajst-10994 45 25 entire entire ADJ ajst-10994 45 26 tag tag NOUN ajst-10994 45 27 sequence sequence NOUN ajst-10994 45 28 after after ADP ajst-10994 45 29 the the DET ajst-10994 45 30 final final ADJ ajst-10994 45 31 transformer transformer NOUN ajst-10994 45 32 layer layer NOUN ajst-10994 45 33 . . PUNCT ajst-10994 46 1 ( ( PUNCT ajst-10994 46 2 2 2 X ajst-10994 46 3 ) ) PUNCT ajst-10994 46 4 taking take VERB ajst-10994 46 5 inspiration inspiration NOUN ajst-10994 46 6 from from ADP ajst-10994 46 7 bert bert PROPN ajst-10994 46 8 's 's PART ajst-10994 46 9 processing processing NOUN ajst-10994 46 10 approach approach NOUN ajst-10994 46 11 , , PUNCT ajst-10994 46 12 manually manually ADV ajst-10994 46 13 introduce introduce VERB ajst-10994 46 14 category category NOUN ajst-10994 46 15 tags tag NOUN ajst-10994 46 16 in in ADP ajst-10994 46 17 the the DET ajst-10994 46 18 token token ADJ ajst-10994 46 19 sequence sequence NOUN ajst-10994 46 20 to to PART ajst-10994 46 21 achieve achieve VERB ajst-10994 46 22 the the DET ajst-10994 46 23 conversion conversion NOUN ajst-10994 46 24 of of ADP ajst-10994 46 25 the the DET ajst-10994 46 26 token token ADJ ajst-10994 46 27 sequence sequence NOUN ajst-10994 46 28 into into ADP ajst-10994 46 29 feature feature NOUN ajst-10994 46 30 maps map NOUN ajst-10994 46 31 . . PUNCT ajst-10994 47 1 in in ADP ajst-10994 47 2 this this DET ajst-10994 47 3 paper paper NOUN ajst-10994 47 4 , , PUNCT ajst-10994 47 5 the the DET ajst-10994 47 6 proposed propose VERB ajst-10994 47 7 tst tst NOUN ajst-10994 47 8 model model NOUN ajst-10994 47 9 adopts adopt VERB ajst-10994 47 10 the the DET ajst-10994 47 11 second second ADJ ajst-10994 47 12 method method NOUN ajst-10994 47 13 to to PART ajst-10994 47 14 derive derive VERB ajst-10994 47 15 the the DET ajst-10994 47 16 feature feature NOUN ajst-10994 47 17 embedding embed VERB ajst-10994 47 18 . . PUNCT ajst-10994 48 1 the the DET ajst-10994 48 2 class class NOUN ajst-10994 48 3 tag tag NOUN ajst-10994 48 4 is be AUX ajst-10994 48 5 actually actually ADV ajst-10994 48 6 a a DET ajst-10994 48 7 erratically erratically ADV ajst-10994 48 8 arranged arrange VERB ajst-10994 48 9 tunable tunable ADJ ajst-10994 48 10 parameter parameter NOUN ajst-10994 48 11 signified signify VERB ajst-10994 48 12 as as ADP ajst-10994 48 13 ๐‘ฅ0 ๐‘ฅ0 NOUN ajst-10994 48 14 โˆˆ โˆˆ PROPN ajst-10994 48 15 r1ร—dim r1ร—dim NOUN ajst-10994 48 16 . . PUNCT ajst-10994 49 1 the the DET ajst-10994 49 2 sequence sequence NOUN ajst-10994 49 3 of of ADP ajst-10994 49 4 tags tag NOUN ajst-10994 49 5 after after SCONJ ajst-10994 49 6 the the DET ajst-10994 49 7 class class NOUN ajst-10994 49 8 tag tag NOUN ajst-10994 49 9 is be AUX ajst-10994 49 10 added add VERB ajst-10994 49 11 is be AUX ajst-10994 49 12 shown show VERB ajst-10994 49 13 in in ADP ajst-10994 49 14 formula formula NOUN ajst-10994 49 15 ( ( PUNCT ajst-10994 49 16 2 2 NUM ajst-10994 49 17 ) ) PUNCT ajst-10994 49 18 . . PUNCT ajst-10994 50 1 1 1 NUM ajst-10994 50 2 2 2 NUM ajst-10994 50 3 3 3 NUM ajst-10994 50 4 0 0 NUM ajst-10994 50 5 ( ( PUNCT ajst-10994 50 6 1 1 X ajst-10994 50 7 ) ) PUNCT ajst-10994 50 8 dim dim NOUN ajst-10994 50 9 [ [ PUNCT ajst-10994 50 10 ; ; PUNCT ajst-10994 50 11 [ [ PUNCT ajst-10994 50 12 , , PUNCT ajst-10994 50 13 , , PUNCT ajst-10994 50 14 , , PUNCT ajst-10994 50 15 ... ... PUNCT ajst-10994 50 16 , , PUNCT ajst-10994 50 17 ] ] PUNCT ajst-10994 51 1 ] ] X ajst-10994 51 2 s s X ajst-10994 51 3 s s AUX ajst-10994 51 4 n n NOUN ajst-10994 51 5 s s NOUN ajst-10994 51 6 s s X ajst-10994 51 7 s s AUX ajst-10994 51 8 s s AUX ajst-10994 51 9 embedding embed VERB ajst-10994 51 10 b b PROPN ajst-10994 51 11 n n PRON ajst-10994 51 12 tokensseq tokensseq NOUN ajst-10994 51 13 x x SYM ajst-10994 51 14 t t NOUN ajst-10994 51 15 t t PROPN ajst-10994 51 16 t t PROPN ajst-10994 51 17 t t PROPN ajst-10994 51 18 w w NOUN ajst-10994 51 19 r r PROPN ajst-10994 51 20 ๏‚ด ๏‚ด NOUN ajst-10994 51 21 ๏€ซ ๏€ซ PROPN ajst-10994 51 22 ๏‚ด ๏‚ด PROPN ajst-10994 51 23 ๏€ฝ ๏€ฝ PROPN ajst-10994 51 24 ๏‚ท ๏‚ท PRON ajst-10994 51 25 ๏ƒŽ ๏ƒŽ NOUN ajst-10994 51 26 ( ( PUNCT ajst-10994 51 27 2 2 NUM ajst-10994 51 28 ) ) PUNCT ajst-10994 51 29 2.1.3 2.1.3 NUM ajst-10994 51 30 . . PUNCT ajst-10994 51 31 location location NOUN ajst-10994 51 32 coding code VERB ajst-10994 51 33 the the DET ajst-10994 51 34 proposed propose VERB ajst-10994 51 35 tst tst NOUN ajst-10994 51 36 ( ( PUNCT ajst-10994 51 37 transformer transformer NOUN ajst-10994 51 38 - - PUNCT ajst-10994 51 39 based base VERB ajst-10994 51 40 time time NOUN ajst-10994 51 41 series series PROPN ajst-10994 51 42 ) ) PUNCT ajst-10994 51 43 model model NOUN ajst-10994 51 44 lacks lack VERB ajst-10994 51 45 filtering filter VERB ajst-10994 51 46 - - PUNCT ajst-10994 51 47 based base VERB ajst-10994 51 48 operations operation NOUN ajst-10994 51 49 and and CCONJ ajst-10994 51 50 does do AUX ajst-10994 51 51 not not PART ajst-10994 51 52 possess possess VERB ajst-10994 51 53 translational translational ADJ ajst-10994 51 54 invariance invariance NOUN ajst-10994 51 55 derivation derivation NOUN ajst-10994 51 56 bias bias NOUN ajst-10994 51 57 . . PUNCT ajst-10994 52 1 furthermore furthermore ADV ajst-10994 52 2 , , PUNCT ajst-10994 52 3 the the DET ajst-10994 52 4 multi multi ADJ ajst-10994 52 5 - - ADJ ajst-10994 52 6 head head ADJ ajst-10994 52 7 , , PUNCT ajst-10994 52 8 self self NOUN ajst-10994 52 9 - - PUNCT ajst-10994 52 10 attention attention NOUN ajst-10994 52 11 mechanism mechanism NOUN ajst-10994 52 12 does do AUX ajst-10994 52 13 not not PART ajst-10994 52 14 include include VERB ajst-10994 52 15 positional positional ADJ ajst-10994 52 16 information information NOUN ajst-10994 52 17 of of ADP ajst-10994 52 18 the the DET ajst-10994 52 19 entries entry NOUN ajst-10994 52 20 during during ADP ajst-10994 52 21 computation computation NOUN ajst-10994 52 22 . . PUNCT ajst-10994 53 1 as as ADP ajst-10994 53 2 a a DET ajst-10994 53 3 result result NOUN ajst-10994 53 4 , , PUNCT ajst-10994 53 5 the the DET ajst-10994 53 6 tst tst NOUN ajst-10994 53 7 model model NOUN ajst-10994 53 8 neglects neglect VERB ajst-10994 53 9 the the DET ajst-10994 53 10 positional positional ADJ ajst-10994 53 11 relationships relationship NOUN ajst-10994 53 12 among among ADP ajst-10994 53 13 various various ADJ ajst-10994 53 14 subsequences subsequence NOUN ajst-10994 53 15 in in ADP ajst-10994 53 16 the the DET ajst-10994 53 17 original original ADJ ajst-10994 53 18 vibration vibration NOUN ajst-10994 53 19 signal signal NOUN ajst-10994 53 20 data datum NOUN ajst-10994 53 21 during during ADP ajst-10994 53 22 the the DET ajst-10994 53 23 computation computation NOUN ajst-10994 53 24 process process NOUN ajst-10994 53 25 . . PUNCT ajst-10994 54 1 to to PART ajst-10994 54 2 address address VERB ajst-10994 54 3 this this DET ajst-10994 54 4 issue issue NOUN ajst-10994 54 5 , , PUNCT ajst-10994 54 6 it it PRON ajst-10994 54 7 is be AUX ajst-10994 54 8 necessary necessary ADJ ajst-10994 54 9 to to PART ajst-10994 54 10 preserve preserve VERB ajst-10994 54 11 the the DET ajst-10994 54 12 absolute absolute ADJ ajst-10994 54 13 and and CCONJ ajst-10994 54 14 relative relative ADJ ajst-10994 54 15 positional positional ADJ ajst-10994 54 16 information information NOUN ajst-10994 54 17 in in ADP ajst-10994 54 18 the the DET ajst-10994 54 19 token token ADJ ajst-10994 54 20 sequence sequence NOUN ajst-10994 54 21 by by ADP ajst-10994 54 22 incorporating incorporate VERB ajst-10994 54 23 the the DET ajst-10994 54 24 method method NOUN ajst-10994 54 25 of of ADP ajst-10994 54 26 position position NOUN ajst-10994 54 27 encoding encoding NOUN ajst-10994 54 28 , , PUNCT ajst-10994 54 29 as as SCONJ ajst-10994 54 30 shown show VERB ajst-10994 54 31 in in ADP ajst-10994 54 32 equation equation NOUN ajst-10994 54 33 ( ( PUNCT ajst-10994 54 34 3 3 NUM ajst-10994 54 35 ) ) PUNCT ajst-10994 54 36 . . PUNCT ajst-10994 55 1 1 1 NUM ajst-10994 55 2 2 2 NUM ajst-10994 55 3 3 3 NUM ajst-10994 55 4 0 0 NUM ajst-10994 55 5 ( ( PUNCT ajst-10994 55 6 1 1 X ajst-10994 55 7 ) ) PUNCT ajst-10994 55 8 dim dim NOUN ajst-10994 55 9 [ [ PUNCT ajst-10994 55 10 ; ; PUNCT ajst-10994 55 11 [ [ PUNCT ajst-10994 55 12 , , PUNCT ajst-10994 55 13 , , PUNCT ajst-10994 55 14 , , PUNCT ajst-10994 55 15 ... ... PUNCT ajst-10994 55 16 , , PUNCT ajst-10994 55 17 ] ] PUNCT ajst-10994 56 1 ] ] X ajst-10994 56 2 s s X ajst-10994 56 3 s s AUX ajst-10994 56 4 n n NOUN ajst-10994 56 5 s s NOUN ajst-10994 56 6 s s NOUN ajst-10994 56 7 s s AUX ajst-10994 56 8 s s AUX ajst-10994 56 9 embedding embed VERB ajst-10994 56 10 pos pos NOUN ajst-10994 56 11 b b PROPN ajst-10994 56 12 n n PRON ajst-10994 56 13 tokensseq tokensseq NOUN ajst-10994 56 14 x x SYM ajst-10994 56 15 t t NOUN ajst-10994 56 16 t t PROPN ajst-10994 56 17 t t PROPN ajst-10994 56 18 t t PROPN ajst-10994 56 19 w w PROPN ajst-10994 56 20 e e NOUN ajst-10994 56 21 r r NOUN ajst-10994 56 22 ๏‚ด ๏‚ด NOUN ajst-10994 56 23 ๏€ซ ๏€ซ PROPN ajst-10994 56 24 ๏‚ด ๏‚ด PROPN ajst-10994 56 25 ๏€ฝ ๏€ฝ PROPN ajst-10994 56 26 ๏‚ท ๏‚ท PRON ajst-10994 56 27 ๏€ซ ๏€ซ ADV ajst-10994 56 28 ๏ƒŽ ๏ƒŽ NOUN ajst-10994 56 29 ( ( PUNCT ajst-10994 56 30 3 3 NUM ajst-10994 56 31 ) ) PUNCT ajst-10994 56 32 where where SCONJ ajst-10994 56 33 epos epos PROPN ajst-10994 56 34 indicates indicate VERB ajst-10994 56 35 the the DET ajst-10994 56 36 location location NOUN ajst-10994 56 37 encoding encoding NOUN ajst-10994 56 38 . . PUNCT ajst-10994 57 1 2.2 2.2 NUM ajst-10994 57 2 . . PUNCT ajst-10994 57 3 transformer transformer NOUN ajst-10994 57 4 layer layer NOUN ajst-10994 57 5 2.2.1 2.2.1 NUM ajst-10994 57 6 . . PUNCT ajst-10994 58 1 multi multi ADJ ajst-10994 58 2 - - ADJ ajst-10994 58 3 head head ADJ ajst-10994 58 4 self self NOUN ajst-10994 58 5 - - PUNCT ajst-10994 58 6 attention attention NOUN ajst-10994 58 7 mechanism mechanism NOUN ajst-10994 58 8 the the DET ajst-10994 58 9 multi multi ADJ ajst-10994 58 10 - - ADJ ajst-10994 58 11 head head ADJ ajst-10994 58 12 self self NOUN ajst-10994 58 13 - - PUNCT ajst-10994 58 14 attention attention NOUN ajst-10994 58 15 mechanism mechanism NOUN ajst-10994 58 16 ( ( PUNCT ajst-10994 58 17 msa msa PROPN ajst-10994 58 18 ) ) PUNCT ajst-10994 58 19 is be AUX ajst-10994 58 20 the the DET ajst-10994 58 21 most most ADV ajst-10994 58 22 crucial crucial ADJ ajst-10994 58 23 component component NOUN ajst-10994 58 24 of of ADP ajst-10994 58 25 the the DET ajst-10994 58 26 transformer transformer NOUN ajst-10994 58 27 model model NOUN ajst-10994 58 28 and and CCONJ ajst-10994 58 29 is be AUX ajst-10994 58 30 entirely entirely ADV ajst-10994 58 31 based base VERB ajst-10994 58 32 on on ADP ajst-10994 58 33 the the DET ajst-10994 58 34 attention attention NOUN ajst-10994 58 35 mechanism mechanism NOUN ajst-10994 58 36 , , PUNCT ajst-10994 58 37 without without ADP ajst-10994 58 38 involving involve VERB ajst-10994 58 39 any any DET ajst-10994 58 40 cnn cnn NOUN ajst-10994 58 41 or or CCONJ ajst-10994 58 42 rnn rnn PROPN ajst-10994 58 43 - - PUNCT ajst-10994 58 44 related relate VERB ajst-10994 58 45 structures structure NOUN ajst-10994 58 46 . . PUNCT ajst-10994 59 1 in in ADP ajst-10994 59 2 general general ADJ ajst-10994 59 3 , , PUNCT ajst-10994 59 4 the the DET ajst-10994 59 5 attention attention NOUN ajst-10994 59 6 mechanism mechanism NOUN ajst-10994 59 7 can can AUX ajst-10994 59 8 be be AUX ajst-10994 59 9 seen see VERB ajst-10994 59 10 as as ADP ajst-10994 59 11 a a DET ajst-10994 59 12 mapping mapping NOUN ajst-10994 59 13 from from ADP ajst-10994 59 14 a a DET ajst-10994 59 15 query query NOUN ajst-10994 59 16 value value NOUN ajst-10994 59 17 ( ( PUNCT ajst-10994 59 18 denoted denote VERB ajst-10994 59 19 as as ADP ajst-10994 59 20 q q NOUN ajst-10994 59 21 ) ) PUNCT ajst-10994 59 22 to to ADP ajst-10994 59 23 a a DET ajst-10994 59 24 set set NOUN ajst-10994 59 25 of of ADP ajst-10994 59 26 key key ADJ ajst-10994 59 27 - - PUNCT ajst-10994 59 28 value value NOUN ajst-10994 59 29 pairs pair NOUN ajst-10994 59 30 ( ( PUNCT ajst-10994 59 31 denoted denote VERB ajst-10994 59 32 as as ADP ajst-10994 59 33 k k PROPN ajst-10994 59 34 and and CCONJ ajst-10994 59 35 v v NOUN ajst-10994 59 36 , , PUNCT ajst-10994 59 37 respectively respectively ADV ajst-10994 59 38 ) ) PUNCT ajst-10994 59 39 , , PUNCT ajst-10994 59 40 where where SCONJ ajst-10994 59 41 q q X ajst-10994 59 42 , , PUNCT ajst-10994 59 43 k k NOUN ajst-10994 59 44 , , PUNCT ajst-10994 59 45 and and CCONJ ajst-10994 59 46 v v NOUN ajst-10994 59 47 are be AUX ajst-10994 59 48 all all PRON ajst-10994 59 49 vectors vector NOUN ajst-10994 59 50 . . PUNCT ajst-10994 60 1 the the DET ajst-10994 60 2 input input NOUN ajst-10994 60 3 to to ADP ajst-10994 60 4 the the DET ajst-10994 60 5 attention attention NOUN ajst-10994 60 6 mechanism mechanism NOUN ajst-10994 60 7 is be AUX ajst-10994 60 8 a a DET ajst-10994 60 9 set set NOUN ajst-10994 60 10 of of ADP ajst-10994 60 11 q q NOUN ajst-10994 60 12 , , PUNCT ajst-10994 60 13 k k NOUN ajst-10994 60 14 , , PUNCT ajst-10994 60 15 and and CCONJ ajst-10994 60 16 v v ADP ajst-10994 60 17 values value NOUN ajst-10994 60 18 , , PUNCT ajst-10994 60 19 and and CCONJ ajst-10994 60 20 the the DET ajst-10994 60 21 output output NOUN ajst-10994 60 22 is be AUX ajst-10994 60 23 the the DET ajst-10994 60 24 weighted weighted ADJ ajst-10994 60 25 sum sum NOUN ajst-10994 60 26 of of ADP ajst-10994 60 27 v v NOUN ajst-10994 60 28 , , PUNCT ajst-10994 60 29 with with ADP ajst-10994 60 30 corresponding correspond VERB ajst-10994 60 31 weights weight NOUN ajst-10994 60 32 represented represent VERB ajst-10994 60 33 by by ADP ajst-10994 60 34 the the DET ajst-10994 60 35 attention attention NOUN ajst-10994 60 36 distribution distribution NOUN ajst-10994 60 37 ( ( PUNCT ajst-10994 60 38 ad ad NOUN ajst-10994 60 39 ) ) PUNCT ajst-10994 60 40 matrix matrix NOUN ajst-10994 60 41 . . PUNCT ajst-10994 61 1 the the DET ajst-10994 61 2 essence essence NOUN ajst-10994 61 3 of of ADP ajst-10994 61 4 the the DET ajst-10994 61 5 ad ad NOUN ajst-10994 61 6 is be AUX ajst-10994 61 7 to to PART ajst-10994 61 8 represent represent VERB ajst-10994 61 9 the the DET ajst-10994 61 10 similarity similarity NOUN ajst-10994 61 11 between between ADP ajst-10994 61 12 the the DET ajst-10994 61 13 two two NUM ajst-10994 61 14 sets set NOUN ajst-10994 61 15 of of ADP ajst-10994 61 16 sequences sequence NOUN ajst-10994 61 17 , , PUNCT ajst-10994 61 18 q q PUNCT ajst-10994 61 19 and and CCONJ ajst-10994 61 20 k. k. PROPN ajst-10994 61 21 there there PRON ajst-10994 61 22 are be VERB ajst-10994 61 23 various various ADJ ajst-10994 61 24 ways way NOUN ajst-10994 61 25 to to PART ajst-10994 61 26 compute compute VERB ajst-10994 61 27 the the DET ajst-10994 61 28 ad ad NOUN ajst-10994 61 29 , , PUNCT ajst-10994 61 30 known know VERB ajst-10994 61 31 as as ADP ajst-10994 61 32 scoring scoring NOUN ajst-10994 61 33 functions function NOUN ajst-10994 61 34 . . PUNCT ajst-10994 62 1 currently currently ADV ajst-10994 62 2 used use VERB ajst-10994 62 3 scoring scoring NOUN ajst-10994 62 4 functions function NOUN ajst-10994 62 5 include include VERB ajst-10994 62 6 the the DET ajst-10994 62 7 scaled scale VERB ajst-10994 62 8 dot dot NOUN ajst-10994 62 9 - - PUNCT ajst-10994 62 10 product product NOUN ajst-10994 62 11 attention attention NOUN ajst-10994 62 12 , , PUNCT ajst-10994 62 13 bahdanau bahdanau ADJ ajst-10994 62 14 attention attention NOUN ajst-10994 62 15 , , PUNCT ajst-10994 62 16 and and CCONJ ajst-10994 62 17 cosine cosine NOUN ajst-10994 62 18 similarity similarity NOUN ajst-10994 62 19 attention attention NOUN ajst-10994 62 20 , , PUNCT ajst-10994 62 21 among among ADP ajst-10994 62 22 81 81 NUM ajst-10994 62 23 others other NOUN ajst-10994 62 24 . . PUNCT ajst-10994 63 1 among among ADP ajst-10994 63 2 these these PRON ajst-10994 63 3 , , PUNCT ajst-10994 63 4 the the DET ajst-10994 63 5 scaled scale VERB ajst-10994 63 6 dot dot NOUN ajst-10994 63 7 - - PUNCT ajst-10994 63 8 product product NOUN ajst-10994 63 9 attention attention NOUN ajst-10994 63 10 is be AUX ajst-10994 63 11 widely widely ADV ajst-10994 63 12 used use VERB ajst-10994 63 13 due due ADP ajst-10994 63 14 to to ADP ajst-10994 63 15 its its PRON ajst-10994 63 16 simplicity simplicity NOUN ajst-10994 63 17 in in ADP ajst-10994 63 18 computation computation NOUN ajst-10994 63 19 , , PUNCT ajst-10994 63 20 ease ease NOUN ajst-10994 63 21 of of ADP ajst-10994 63 22 parallelization parallelization NOUN ajst-10994 63 23 , , PUNCT ajst-10994 63 24 and and CCONJ ajst-10994 63 25 no no DET ajst-10994 63 26 introduction introduction NOUN ajst-10994 63 27 of of ADP ajst-10994 63 28 additional additional ADJ ajst-10994 63 29 parameters parameter NOUN ajst-10994 63 30 into into ADP ajst-10994 63 31 the the DET ajst-10994 63 32 model model NOUN ajst-10994 63 33 . . PUNCT ajst-10994 64 1 the the DET ajst-10994 64 2 specific specific ADJ ajst-10994 64 3 computation computation NOUN ajst-10994 64 4 of of ADP ajst-10994 64 5 the the DET ajst-10994 64 6 scaled scale VERB ajst-10994 64 7 dotproduct dotproduct NOUN ajst-10994 64 8 attention attention NOUN ajst-10994 64 9 is be AUX ajst-10994 64 10 as as SCONJ ajst-10994 64 11 follows follow VERB ajst-10994 64 12 . . PUNCT ajst-10994 65 1 ( ( PUNCT ajst-10994 65 2 , , PUNCT ajst-10994 65 3 , , PUNCT ajst-10994 65 4 ) ) PUNCT ajst-10994 65 5 max max PROPN ajst-10994 65 6 ( ( PUNCT ajst-10994 65 7 ) ) PUNCT ajst-10994 65 8 t t PROPN ajst-10994 66 1 k k PROPN ajst-10994 66 2 qk qk ADP ajst-10994 66 3 attention attention NOUN ajst-10994 66 4 q q PROPN ajst-10994 67 1 k k X ajst-10994 67 2 v v X ajst-10994 67 3 soft soft ADJ ajst-10994 67 4 v v NOUN ajst-10994 67 5 d d X ajst-10994 67 6 ๏€ฝ ๏€ฝ PROPN ajst-10994 67 7 ( ( PUNCT ajst-10994 67 8 4 4 NUM ajst-10994 67 9 ) ) PUNCT ajst-10994 67 10 where where SCONJ ajst-10994 67 11 dk dk PROPN ajst-10994 67 12 indicates indicate VERB ajst-10994 67 13 the the DET ajst-10994 67 14 entered enter VERB ajst-10994 67 15 dimensions dimension NOUN ajst-10994 67 16 of of ADP ajst-10994 67 17 q q PROPN ajst-10994 67 18 and and CCONJ ajst-10994 67 19 k. k. PROPN ajst-10994 67 20 2.2.2 2.2.2 NUM ajst-10994 67 21 . . PUNCT ajst-10994 68 1 multilayer multilayer PROPN ajst-10994 68 2 perceptron perceptron PROPN ajst-10994 68 3 module module NOUN ajst-10994 68 4 to to PART ajst-10994 68 5 enable enable VERB ajst-10994 68 6 the the DET ajst-10994 68 7 proposed propose VERB ajst-10994 68 8 tst tst NOUN ajst-10994 68 9 ( ( PUNCT ajst-10994 68 10 transformer transformer NOUN ajst-10994 68 11 - - PUNCT ajst-10994 68 12 based base VERB ajst-10994 68 13 time time NOUN ajst-10994 68 14 series series PROPN ajst-10994 68 15 ) ) PUNCT ajst-10994 68 16 model model NOUN ajst-10994 68 17 to to PART ajst-10994 68 18 achieve achieve VERB ajst-10994 68 19 more more ADV ajst-10994 68 20 complex complex ADJ ajst-10994 68 21 non non ADJ ajst-10994 68 22 - - ADJ ajst-10994 68 23 linear linear ADJ ajst-10994 68 24 mappings mapping NOUN ajst-10994 68 25 , , PUNCT ajst-10994 68 26 each each DET ajst-10994 68 27 transformer transformer NOUN ajst-10994 68 28 module module NOUN ajst-10994 68 29 also also ADV ajst-10994 68 30 includes include VERB ajst-10994 68 31 a a DET ajst-10994 68 32 multilayer multilayer ADJ ajst-10994 68 33 perceptron perceptron PROPN ajst-10994 68 34 block block NOUN ajst-10994 68 35 ( ( PUNCT ajst-10994 68 36 mlp mlp PROPN ajst-10994 68 37 ) ) PUNCT ajst-10994 68 38 . . PUNCT ajst-10994 69 1 the the DET ajst-10994 69 2 mlp mlp NOUN ajst-10994 69 3 module module NOUN ajst-10994 69 4 within within ADP ajst-10994 69 5 the the DET ajst-10994 69 6 l l NOUN ajst-10994 69 7 - - PUNCT ajst-10994 69 8 th th VERB ajst-10994 69 9 transformer transformer NOUN ajst-10994 69 10 module module NOUN ajst-10994 69 11 consists consist VERB ajst-10994 69 12 of of ADP ajst-10994 69 13 a a DET ajst-10994 69 14 non non ADJ ajst-10994 69 15 - - ADJ ajst-10994 69 16 linear linear ADJ ajst-10994 69 17 mapping mapping NOUN ajst-10994 69 18 layer layer NOUN ajst-10994 69 19 with with ADP ajst-10994 69 20 an an DET ajst-10994 69 21 activation activation NOUN ajst-10994 69 22 function function NOUN ajst-10994 69 23 and and CCONJ ajst-10994 69 24 a a DET ajst-10994 69 25 linear linear ADJ ajst-10994 69 26 mapping mapping NOUN ajst-10994 69 27 layer layer NOUN ajst-10994 69 28 , , PUNCT ajst-10994 69 29 as as SCONJ ajst-10994 69 30 shown show VERB ajst-10994 69 31 in in ADP ajst-10994 69 32 equation equation NOUN ajst-10994 69 33 ( ( PUNCT ajst-10994 69 34 5 5 NUM ajst-10994 69 35 ) ) PUNCT ajst-10994 69 36 . . PUNCT ajst-10994 70 1 1 1 NUM ajst-10994 70 2 1 1 NUM ajst-10994 70 3 2 2 NUM ajst-10994 70 4 2 2 NUM ajst-10994 70 5 ( ( PUNCT ajst-10994 70 6 ) ) PUNCT ajst-10994 70 7 ( ( PUNCT ajst-10994 70 8 ) ) PUNCT ajst-10994 70 9 msa msa PROPN ajst-10994 70 10 msa msa PROPN ajst-10994 70 11 l l PROPN ajst-10994 70 12 l l NOUN ajst-10994 70 13 l l NOUN ajst-10994 70 14 l l NOUN ajst-10994 70 15 l l NOUN ajst-10994 70 16 lmsp lmsp NOUN ajst-10994 70 17 y y PROPN ajst-10994 70 18 activation activation NOUN ajst-10994 70 19 y y PROPN ajst-10994 70 20 w w PROPN ajst-10994 70 21 b b PROPN ajst-10994 70 22 w w PROPN ajst-10994 70 23 b๏€ฝ b๏€ฝ PROPN ajst-10994 70 24 ๏€ซ ๏€ซ PROPN ajst-10994 70 25 ๏€ซ ๏€ซ X ajst-10994 70 26 ( ( PUNCT ajst-10994 70 27 5 5 X ajst-10994 70 28 ) ) PUNCT ajst-10994 70 29 2.2.3 2.2.3 NUM ajst-10994 70 30 . . PUNCT ajst-10994 71 1 classifier classifier NOUN ajst-10994 71 2 layer layer NOUN ajst-10994 71 3 the the DET ajst-10994 71 4 classifier classifier NOUN ajst-10994 71 5 layer layer NOUN ajst-10994 71 6 is be AUX ajst-10994 71 7 the the DET ajst-10994 71 8 final final ADJ ajst-10994 71 9 layer layer NOUN ajst-10994 71 10 of of ADP ajst-10994 71 11 the the DET ajst-10994 71 12 tst tst NOUN ajst-10994 71 13 ( ( PUNCT ajst-10994 71 14 transformer transformer NOUN ajst-10994 71 15 - - PUNCT ajst-10994 71 16 based base VERB ajst-10994 71 17 time time NOUN ajst-10994 71 18 series series PROPN ajst-10994 71 19 ) ) PUNCT ajst-10994 71 20 model model NOUN ajst-10994 71 21 in in ADP ajst-10994 71 22 this this DET ajst-10994 71 23 paper paper NOUN ajst-10994 71 24 , , PUNCT ajst-10994 71 25 and and CCONJ ajst-10994 71 26 it it PRON ajst-10994 71 27 is be AUX ajst-10994 71 28 responsible responsible ADJ ajst-10994 71 29 for for ADP ajst-10994 71 30 converting convert VERB ajst-10994 71 31 the the DET ajst-10994 71 32 feature feature NOUN ajst-10994 71 33 maps map NOUN ajst-10994 71 34 extracted extract VERB ajst-10994 71 35 by by ADP ajst-10994 71 36 the the DET ajst-10994 71 37 tst tst NOUN ajst-10994 71 38 model model NOUN ajst-10994 71 39 into into ADP ajst-10994 71 40 one one NUM ajst-10994 71 41 - - PUNCT ajst-10994 71 42 hot hot ADJ ajst-10994 71 43 encoded encode VERB ajst-10994 71 44 representations representation NOUN ajst-10994 71 45 suitable suitable ADJ ajst-10994 71 46 for for ADP ajst-10994 71 47 direct direct ADJ ajst-10994 71 48 fault fault NOUN ajst-10994 71 49 diagnosis diagnosis NOUN ajst-10994 71 50 . . PUNCT ajst-10994 72 1 its its PRON ajst-10994 72 2 basic basic ADJ ajst-10994 72 3 structure structure NOUN ajst-10994 72 4 is be AUX ajst-10994 72 5 similar similar ADJ ajst-10994 72 6 to to ADP ajst-10994 72 7 that that PRON ajst-10994 72 8 of of ADP ajst-10994 72 9 the the DET ajst-10994 72 10 mlp mlp NOUN ajst-10994 72 11 module module NOUN ajst-10994 72 12 , , PUNCT ajst-10994 72 13 as as SCONJ ajst-10994 72 14 shown show VERB ajst-10994 72 15 in in ADP ajst-10994 72 16 equation equation NOUN ajst-10994 72 17 ( ( PUNCT ajst-10994 72 18 6 6 NUM ajst-10994 72 19 ) ) PUNCT ajst-10994 72 20 . . PUNCT ajst-10994 73 1 ( ( PUNCT ajst-10994 73 2 ) ) PUNCT ajst-10994 73 3 max max PROPN ajst-10994 73 4 ( ( PUNCT ajst-10994 73 5 ) ) PUNCT ajst-10994 73 6 class class NOUN ajst-10994 73 7 class class NOUN ajst-10994 73 8 classsclasslayer classsclasslayer PROPN ajst-10994 73 9 y y PROPN ajst-10994 73 10 y y PROPN ajst-10994 73 11 soft soft PROPN ajst-10994 73 12 yw yw PROPN ajst-10994 73 13 b๏€ฝ b๏€ฝ PROPN ajst-10994 73 14 ๏€ฝ ๏€ฝ PROPN ajst-10994 73 15 ๏€ซ ๏€ซ PUNCT ajst-10994 73 16 ( ( PUNCT ajst-10994 73 17 6 6 NUM ajst-10994 73 18 ) ) SYM ajst-10994 73 19 3 3 NUM ajst-10994 73 20 . . PUNCT ajst-10994 73 21 data datum NOUN ajst-10994 73 22 sets set NOUN ajst-10994 73 23 and and CCONJ ajst-10994 73 24 data datum NOUN ajst-10994 73 25 preprocessing preprocessing NOUN ajst-10994 73 26 methods method NOUN ajst-10994 73 27 in in ADP ajst-10994 73 28 this this DET ajst-10994 73 29 paper paper NOUN ajst-10994 73 30 , , PUNCT ajst-10994 73 31 we we PRON ajst-10994 73 32 utilized utilize VERB ajst-10994 73 33 publicly publicly ADV ajst-10994 73 34 available available ADJ ajst-10994 73 35 experimental experimental ADJ ajst-10994 73 36 datasets dataset NOUN ajst-10994 73 37 from from ADP ajst-10994 73 38 the the DET ajst-10994 73 39 field field NOUN ajst-10994 73 40 of of ADP ajst-10994 73 41 rotating rotate VERB ajst-10994 73 42 machinery machinery NOUN ajst-10994 73 43 fault fault NOUN ajst-10994 73 44 diagnosis diagnosis NOUN ajst-10994 73 45 to to PART ajst-10994 73 46 train train VERB ajst-10994 73 47 , , PUNCT ajst-10994 73 48 test test NOUN ajst-10994 73 49 , , PUNCT ajst-10994 73 50 and and CCONJ ajst-10994 73 51 validate validate VERB ajst-10994 73 52 the the DET ajst-10994 73 53 proposed propose VERB ajst-10994 73 54 tst tst NOUN ajst-10994 73 55 ( ( PUNCT ajst-10994 73 56 transformer transformer NOUN ajst-10994 73 57 - - PUNCT ajst-10994 73 58 based base VERB ajst-10994 73 59 time time NOUN ajst-10994 73 60 series series PROPN ajst-10994 73 61 ) ) PUNCT ajst-10994 73 62 model model NOUN ajst-10994 73 63 . . PUNCT ajst-10994 74 1 the the DET ajst-10994 74 2 experimental experimental ADJ ajst-10994 74 3 datasets dataset NOUN ajst-10994 74 4 used use VERB ajst-10994 74 5 include include VERB ajst-10994 74 6 the the DET ajst-10994 74 7 case case NOUN ajst-10994 74 8 western western PROPN ajst-10994 74 9 reserve reserve PROPN ajst-10994 74 10 university university PROPN ajst-10994 74 11 ( ( PUNCT ajst-10994 74 12 cwru cwru PROPN ajst-10994 74 13 ) ) PUNCT ajst-10994 74 14 rolling roll VERB ajst-10994 74 15 bearing bear VERB ajst-10994 74 16 fault fault NOUN ajst-10994 74 17 dataset dataset VERB ajst-10994 74 18 , , PUNCT ajst-10994 74 19 the the DET ajst-10994 74 20 xi'an xi'an PROPN ajst-10994 74 21 jiaotong jiaotong PROPN ajst-10994 74 22 university university PROPN ajst-10994 74 23 ( ( PUNCT ajst-10994 74 24 xjtu xjtu ADV ajst-10994 74 25 ) ) PUNCT ajst-10994 74 26 rolling roll VERB ajst-10994 74 27 bearing bear VERB ajst-10994 74 28 fatigue fatigue NOUN ajst-10994 74 29 fault fault NOUN ajst-10994 74 30 dataset dataset VERB ajst-10994 74 31 , , PUNCT ajst-10994 74 32 and and CCONJ ajst-10994 74 33 the the DET ajst-10994 74 34 university university NOUN ajst-10994 74 35 of of ADP ajst-10994 74 36 connecticut connecticut PROPN ajst-10994 74 37 ( ( PUNCT ajst-10994 74 38 uconn uconn NOUN ajst-10994 74 39 ) ) PUNCT ajst-10994 74 40 gearbox gearbox NOUN ajst-10994 74 41 fault fault NOUN ajst-10994 74 42 dataset dataset VERB ajst-10994 74 43 . . PUNCT ajst-10994 75 1 these these DET ajst-10994 75 2 datasets dataset NOUN ajst-10994 75 3 consist consist VERB ajst-10994 75 4 of of ADP ajst-10994 75 5 three three NUM ajst-10994 75 6 typical typical ADJ ajst-10994 75 7 forms form NOUN ajst-10994 75 8 of of ADP ajst-10994 75 9 rotating rotate VERB ajst-10994 75 10 machinery machinery NOUN ajst-10994 75 11 faults fault NOUN ajst-10994 75 12 : : PUNCT ajst-10994 75 13 pre pre ADJ ajst-10994 75 14 - - ADJ ajst-10994 75 15 fault fault VERB ajst-10994 75 16 rolling rolling ADJ ajst-10994 75 17 bearing bearing NOUN ajst-10994 75 18 , , PUNCT ajst-10994 75 19 rolling rolling ADJ ajst-10994 75 20 bearing bear VERB ajst-10994 75 21 fatigue fatigue NOUN ajst-10994 75 22 fault fault NOUN ajst-10994 75 23 , , PUNCT ajst-10994 75 24 and and CCONJ ajst-10994 75 25 gearbox gearbox ADJ ajst-10994 75 26 fault fault NOUN ajst-10994 75 27 . . PUNCT ajst-10994 76 1 the the DET ajst-10994 76 2 experimental experimental ADJ ajst-10994 76 3 equipment equipment NOUN ajst-10994 76 4 used use VERB ajst-10994 76 5 for for ADP ajst-10994 76 6 data data NOUN ajst-10994 76 7 collection collection NOUN ajst-10994 76 8 for for ADP ajst-10994 76 9 each each DET ajst-10994 76 10 dataset dataset NOUN ajst-10994 76 11 is be AUX ajst-10994 76 12 shown show VERB ajst-10994 76 13 in in ADP ajst-10994 76 14 figure figure NOUN ajst-10994 76 15 2 2 NUM ajst-10994 76 16 . . PUNCT ajst-10994 77 1 below below ADV ajst-10994 77 2 is be AUX ajst-10994 77 3 an an DET ajst-10994 77 4 introductory introductory ADJ ajst-10994 77 5 glimpse glimpse NOUN ajst-10994 77 6 into into ADP ajst-10994 77 7 the the DET ajst-10994 77 8 datasets dataset NOUN ajst-10994 77 9 and and CCONJ ajst-10994 77 10 the the DET ajst-10994 77 11 data data NOUN ajst-10994 77 12 conditioning conditioning NOUN ajst-10994 77 13 methods method NOUN ajst-10994 77 14 used use VERB ajst-10994 77 15 . . PUNCT ajst-10994 78 1 figure figure NOUN ajst-10994 78 2 2 2 NUM ajst-10994 78 3 . . PUNCT ajst-10994 78 4 schematic schematic ADJ ajst-10994 78 5 diagram diagram NOUN ajst-10994 78 6 of of ADP ajst-10994 78 7 experimental experimental ADJ ajst-10994 78 8 equipment equipment NOUN ajst-10994 78 9 during during ADP ajst-10994 78 10 data data NOUN ajst-10994 78 11 collection collection NOUN ajst-10994 78 12 of of ADP ajst-10994 78 13 each each DET ajst-10994 78 14 experimental experimental ADJ ajst-10994 78 15 data datum NOUN ajst-10994 79 1 set set VERB ajst-10994 79 2 ( ( PUNCT ajst-10994 79 3 a a PRON ajst-10994 79 4 ) ) PUNCT ajst-10994 79 5 cwru cwru PROPN ajst-10994 79 6 data datum NOUN ajst-10994 80 1 set set NOUN ajst-10994 80 2 ( ( PUNCT ajst-10994 80 3 b b NOUN ajst-10994 80 4 ) ) PUNCT ajst-10994 80 5 xjtu xjtu PROPN ajst-10994 81 1 data data PROPN ajst-10994 81 2 set set VERB ajst-10994 81 3 ( ( PUNCT ajst-10994 81 4 c c NOUN ajst-10994 81 5 ) ) PUNCT ajst-10994 81 6 ucoon ucoon NOUN ajst-10994 81 7 data datum NOUN ajst-10994 81 8 set set VERB ajst-10994 81 9 3.1 3.1 NUM ajst-10994 81 10 . . PUNCT ajst-10994 82 1 cwru cwru PROPN ajst-10994 82 2 data datum NOUN ajst-10994 82 3 set set VERB ajst-10994 82 4 the the DET ajst-10994 82 5 cwru cwru NOUN ajst-10994 82 6 dataset dataset NOUN ajst-10994 82 7 was be AUX ajst-10994 82 8 collected collect VERB ajst-10994 82 9 with with ADP ajst-10994 82 10 a a DET ajst-10994 82 11 high high ADJ ajst-10994 82 12 sampling sample VERB ajst-10994 82 13 frequency frequency NOUN ajst-10994 82 14 of of ADP ajst-10994 82 15 1.2khz 1.2khz NUM ajst-10994 82 16 and and CCONJ ajst-10994 82 17 4.8khz 4.8khz NUM ajst-10994 82 18 during during ADP ajst-10994 82 19 the the DET ajst-10994 82 20 experimental experimental ADJ ajst-10994 82 21 process[7 process[7 NOUN ajst-10994 82 22 ] ] PUNCT ajst-10994 82 23 . . PUNCT ajst-10994 83 1 as as ADP ajst-10994 83 2 a a DET ajst-10994 83 3 result result NOUN ajst-10994 83 4 , , PUNCT ajst-10994 83 5 the the DET ajst-10994 83 6 original original ADJ ajst-10994 83 7 vibration vibration NOUN ajst-10994 83 8 signal signal NOUN ajst-10994 83 9 data datum NOUN ajst-10994 83 10 for for ADP ajst-10994 83 11 each each DET ajst-10994 83 12 faulty faulty ADJ ajst-10994 83 13 bearing bearing NOUN ajst-10994 83 14 contains contain VERB ajst-10994 83 15 tens ten NOUN ajst-10994 83 16 of of ADP ajst-10994 83 17 thousands thousand NOUN ajst-10994 83 18 of of ADP ajst-10994 83 19 data datum NOUN ajst-10994 83 20 points point NOUN ajst-10994 83 21 . . PUNCT ajst-10994 84 1 due due ADP ajst-10994 84 2 to to ADP ajst-10994 84 3 their their PRON ajst-10994 84 4 excessive excessive ADJ ajst-10994 84 5 length length NOUN ajst-10994 84 6 , , PUNCT ajst-10994 84 7 these these DET ajst-10994 84 8 data datum NOUN ajst-10994 84 9 can can AUX ajst-10994 84 10 not not PART ajst-10994 84 11 be be AUX ajst-10994 84 12 directly directly ADV ajst-10994 84 13 used use VERB ajst-10994 84 14 for for ADP ajst-10994 84 15 training training NOUN ajst-10994 84 16 and and CCONJ ajst-10994 84 17 testing test VERB ajst-10994 84 18 the the DET ajst-10994 84 19 model model NOUN ajst-10994 84 20 . . PUNCT ajst-10994 85 1 therefore therefore ADV ajst-10994 85 2 , , PUNCT ajst-10994 85 3 a a DET ajst-10994 85 4 resampling resample VERB ajst-10994 85 5 method method NOUN ajst-10994 85 6 is be AUX ajst-10994 85 7 employed employ VERB ajst-10994 85 8 to to PART ajst-10994 85 9 preprocess preprocess VERB ajst-10994 85 10 the the DET ajst-10994 85 11 data datum NOUN ajst-10994 85 12 into into ADP ajst-10994 85 13 a a DET ajst-10994 85 14 specified specified ADJ ajst-10994 85 15 length length NOUN ajst-10994 85 16 of of ADP ajst-10994 85 17 2048 2048 NUM ajst-10994 85 18 data datum NOUN ajst-10994 85 19 points point NOUN ajst-10994 85 20 . . PUNCT ajst-10994 86 1 the the DET ajst-10994 86 2 resampling resample VERB ajst-10994 86 3 method method NOUN ajst-10994 86 4 used use VERB ajst-10994 86 5 in in ADP ajst-10994 86 6 this this DET ajst-10994 86 7 chapter chapter NOUN ajst-10994 86 8 is be AUX ajst-10994 86 9 illustrated illustrate VERB ajst-10994 86 10 in in ADP ajst-10994 86 11 figure figure NOUN ajst-10994 86 12 3 3 NUM ajst-10994 86 13 and and CCONJ ajst-10994 86 14 mainly mainly ADV ajst-10994 86 15 involves involve VERB ajst-10994 86 16 two two NUM ajst-10994 86 17 sampling sample VERB ajst-10994 86 18 approaches approach NOUN ajst-10994 86 19 . . PUNCT ajst-10994 87 1 the the DET ajst-10994 87 2 first first ADJ ajst-10994 87 3 approach approach NOUN ajst-10994 87 4 is be AUX ajst-10994 87 5 to to PART ajst-10994 87 6 resample resample VERB ajst-10994 87 7 directly directly ADV ajst-10994 87 8 without without ADP ajst-10994 87 9 any any DET ajst-10994 87 10 overlap overlap NOUN ajst-10994 87 11 in in ADP ajst-10994 87 12 the the DET ajst-10994 87 13 original original ADJ ajst-10994 87 14 vibration vibration NOUN ajst-10994 87 15 signal signal NOUN ajst-10994 87 16 data datum NOUN ajst-10994 87 17 , , PUNCT ajst-10994 87 18 which which PRON ajst-10994 87 19 is be AUX ajst-10994 87 20 the the DET ajst-10994 87 21 most most ADV ajst-10994 87 22 common common ADJ ajst-10994 87 23 preprocessing preprocessing NOUN ajst-10994 87 24 method method NOUN ajst-10994 87 25 . . PUNCT ajst-10994 88 1 the the DET ajst-10994 88 2 second second ADJ ajst-10994 88 3 approach approach NOUN ajst-10994 88 4 is be AUX ajst-10994 88 5 a a DET ajst-10994 88 6 special special ADJ ajst-10994 88 7 resampling resampling NOUN ajst-10994 88 8 method method NOUN ajst-10994 88 9 where where SCONJ ajst-10994 88 10 overlapping overlap VERB ajst-10994 88 11 regions region NOUN ajst-10994 88 12 are be AUX ajst-10994 88 13 deliberately deliberately ADV ajst-10994 88 14 set set VERB ajst-10994 88 15 during during ADP ajst-10994 88 16 the the DET ajst-10994 88 17 sampling sampling NOUN ajst-10994 88 18 process process NOUN ajst-10994 88 19 , , PUNCT ajst-10994 88 20 providing provide VERB ajst-10994 88 21 a a DET ajst-10994 88 22 form form NOUN ajst-10994 88 23 of of ADP ajst-10994 88 24 data datum NOUN ajst-10994 88 25 augmentation augmentation NOUN ajst-10994 88 26 . . PUNCT ajst-10994 89 1 based base VERB ajst-10994 89 2 on on ADP ajst-10994 89 3 the the DET ajst-10994 89 4 original original ADJ ajst-10994 89 5 data datum NOUN ajst-10994 89 6 from from ADP ajst-10994 89 7 the the DET ajst-10994 89 8 cwru cwru NOUN ajst-10994 89 9 dataset dataset NOUN ajst-10994 89 10 and and CCONJ ajst-10994 89 11 the the DET ajst-10994 89 12 resampling resample VERB ajst-10994 89 13 approaches approach NOUN ajst-10994 89 14 depicted depict VERB ajst-10994 89 15 in in ADP ajst-10994 89 16 figures figure NOUN ajst-10994 89 17 2 2 NUM ajst-10994 89 18 - - SYM ajst-10994 89 19 3 3 NUM ajst-10994 89 20 , , PUNCT ajst-10994 89 21 we we PRON ajst-10994 89 22 constructed construct VERB ajst-10994 89 23 a a DET ajst-10994 89 24 cwru cwru NOUN ajst-10994 89 25 dataset dataset NOUN ajst-10994 89 26 containing contain VERB ajst-10994 89 27 9000 9000 NUM ajst-10994 89 28 samples sample NOUN ajst-10994 89 29 . . PUNCT ajst-10994 90 1 among among ADP ajst-10994 90 2 these these PRON ajst-10994 90 3 , , PUNCT ajst-10994 90 4 7000 7000 NUM ajst-10994 90 5 samples sample NOUN ajst-10994 90 6 form form VERB ajst-10994 90 7 the the DET ajst-10994 90 8 training training NOUN ajst-10994 90 9 set set NOUN ajst-10994 90 10 , , PUNCT ajst-10994 90 11 used use VERB ajst-10994 90 12 to to PART ajst-10994 90 13 train train VERB ajst-10994 90 14 the the DET ajst-10994 90 15 tst tst NOUN ajst-10994 90 16 model model NOUN ajst-10994 90 17 proposed propose VERB ajst-10994 90 18 in in ADP ajst-10994 90 19 this this DET ajst-10994 90 20 chapter chapter NOUN ajst-10994 90 21 . . PUNCT ajst-10994 91 1 the the DET ajst-10994 91 2 remaining remain VERB ajst-10994 91 3 2000 2000 NUM ajst-10994 91 4 samples sample NOUN ajst-10994 91 5 compose compose VERB ajst-10994 91 6 the the DET ajst-10994 91 7 test test NOUN ajst-10994 91 8 set set NOUN ajst-10994 91 9 , , PUNCT ajst-10994 91 10 utilized utilize VERB ajst-10994 91 11 to to PART ajst-10994 91 12 evaluate evaluate VERB ajst-10994 91 13 the the DET ajst-10994 91 14 fault fault NOUN ajst-10994 91 15 diagnosis diagnosis NOUN ajst-10994 91 16 performance performance NOUN ajst-10994 91 17 of of ADP ajst-10994 91 18 the the DET ajst-10994 91 19 tst tst NOUN ajst-10994 91 20 model model NOUN ajst-10994 91 21 . . PUNCT ajst-10994 92 1 there there PRON ajst-10994 92 2 is be VERB ajst-10994 92 3 no no DET ajst-10994 92 4 overlap overlap NOUN ajst-10994 92 5 between between ADP ajst-10994 92 6 the the DET ajst-10994 92 7 training training NOUN ajst-10994 92 8 and and CCONJ ajst-10994 92 9 test test NOUN ajst-10994 92 10 sets set NOUN ajst-10994 92 11 . . PUNCT ajst-10994 93 1 figure figure NOUN ajst-10994 93 2 3 3 NUM ajst-10994 93 3 . . PUNCT ajst-10994 93 4 data datum NOUN ajst-10994 93 5 set set VERB ajst-10994 93 6 resampling resample VERB ajst-10994 93 7 82 82 NUM ajst-10994 93 8 3.2 3.2 NUM ajst-10994 93 9 . . PUNCT ajst-10994 94 1 xjtu xjtu PROPN ajst-10994 94 2 data data PROPN ajst-10994 94 3 set set VERB ajst-10994 94 4 the the DET ajst-10994 94 5 xjtu xjtu PROPN ajst-10994 94 6 dataset dataset PROPN ajst-10994 94 7 is be AUX ajst-10994 94 8 another another PRON ajst-10994 94 9 widely widely ADV ajst-10994 94 10 used use VERB ajst-10994 94 11 rolling rolling NOUN ajst-10994 94 12 bearing bear VERB ajst-10994 94 13 fault fault NOUN ajst-10994 94 14 dataset dataset NOUN ajst-10994 94 15 , , PUNCT ajst-10994 94 16 collected collect VERB ajst-10994 94 17 and and CCONJ ajst-10994 94 18 organized organize VERB ajst-10994 94 19 by by ADP ajst-10994 94 20 researchers researcher NOUN ajst-10994 94 21 from from ADP ajst-10994 94 22 the the DET ajst-10994 94 23 institute institute NOUN ajst-10994 94 24 of of ADP ajst-10994 94 25 design design NOUN ajst-10994 94 26 science science NOUN ajst-10994 94 27 and and CCONJ ajst-10994 94 28 basic basic ADJ ajst-10994 94 29 components component NOUN ajst-10994 94 30 at at ADP ajst-10994 94 31 xi'an xi'an PROPN ajst-10994 94 32 jiaotong jiaotong PROPN ajst-10994 94 33 university university PROPN ajst-10994 94 34 . . PUNCT ajst-10994 95 1 unlike unlike ADP ajst-10994 95 2 the the DET ajst-10994 95 3 cwru cwru PROPN ajst-10994 95 4 dataset dataset PROPN ajst-10994 95 5 , , PUNCT ajst-10994 95 6 the the DET ajst-10994 95 7 xjtu xjtu PROPN ajst-10994 95 8 dataset dataset VERB ajst-10994 95 9 consists consist NOUN ajst-10994 95 10 of of ADP ajst-10994 95 11 the the DET ajst-10994 95 12 entire entire ADJ ajst-10994 95 13 process process NOUN ajst-10994 95 14 of of ADP ajst-10994 95 15 15 15 NUM ajst-10994 95 16 bearings bearing NOUN ajst-10994 95 17 from from ADP ajst-10994 95 18 normal normal ADJ ajst-10994 95 19 operation operation NOUN ajst-10994 95 20 to to ADP ajst-10994 95 21 fatigue fatigue NOUN ajst-10994 95 22 failure failure NOUN ajst-10994 95 23 , , PUNCT ajst-10994 95 24 making make VERB ajst-10994 95 25 it it PRON ajst-10994 95 26 a a DET ajst-10994 95 27 rolling rolling ADJ ajst-10994 95 28 bearing bear VERB ajst-10994 95 29 fatigue fatigue NOUN ajst-10994 95 30 fault fault NOUN ajst-10994 95 31 dataset dataset VERB ajst-10994 95 32 . . PUNCT ajst-10994 96 1 the the DET ajst-10994 96 2 experimental experimental ADJ ajst-10994 96 3 setup setup NOUN ajst-10994 96 4 for for ADP ajst-10994 96 5 the the DET ajst-10994 96 6 xjtu xjtu PROPN ajst-10994 96 7 dataset dataset PROPN ajst-10994 96 8 includes include VERB ajst-10994 96 9 an an DET ajst-10994 96 10 alternating alternate VERB ajst-10994 96 11 current current ADJ ajst-10994 96 12 motor motor NOUN ajst-10994 96 13 , , PUNCT ajst-10994 96 14 motor motor NOUN ajst-10994 96 15 speed speed NOUN ajst-10994 96 16 controller controller NOUN ajst-10994 96 17 , , PUNCT ajst-10994 96 18 support support NOUN ajst-10994 96 19 shaft shaft NOUN ajst-10994 96 20 , , PUNCT ajst-10994 96 21 and and CCONJ ajst-10994 96 22 hydraulic hydraulic ADJ ajst-10994 96 23 loading loading NOUN ajst-10994 96 24 system[8 system[8 NOUN ajst-10994 96 25 ] ] PUNCT ajst-10994 96 26 . . PUNCT ajst-10994 97 1 the the DET ajst-10994 97 2 vibration vibration NOUN ajst-10994 97 3 signals signal NOUN ajst-10994 97 4 of of ADP ajst-10994 97 5 the the DET ajst-10994 97 6 tested test VERB ajst-10994 97 7 bearings bearing NOUN ajst-10994 97 8 are be AUX ajst-10994 97 9 measured measure VERB ajst-10994 97 10 using use VERB ajst-10994 97 11 accelerometers accelerometer NOUN ajst-10994 97 12 installed instal VERB ajst-10994 97 13 in in ADP ajst-10994 97 14 both both CCONJ ajst-10994 97 15 horizontal horizontal ADJ ajst-10994 97 16 and and CCONJ ajst-10994 97 17 vertical vertical ADJ ajst-10994 97 18 directions direction NOUN ajst-10994 97 19 . . PUNCT ajst-10994 98 1 the the DET ajst-10994 98 2 xjtu xjtu PROPN ajst-10994 98 3 dataset dataset PROPN ajst-10994 98 4 contains contain VERB ajst-10994 98 5 fatigue fatigue NOUN ajst-10994 98 6 fault fault NOUN ajst-10994 98 7 data datum NOUN ajst-10994 98 8 from from ADP ajst-10994 98 9 15 15 NUM ajst-10994 98 10 ldk ldk PROPN ajst-10994 98 11 uer204 uer204 PROPN ajst-10994 98 12 model model NOUN ajst-10994 98 13 rolling rolling ADJ ajst-10994 98 14 bearings bearing NOUN ajst-10994 98 15 . . PUNCT ajst-10994 99 1 in in ADP ajst-10994 99 2 this this DET ajst-10994 99 3 paper paper NOUN ajst-10994 99 4 , , PUNCT ajst-10994 99 5 we we PRON ajst-10994 99 6 selected select VERB ajst-10994 99 7 bearings bearing NOUN ajst-10994 99 8 3_1 3_1 NUM ajst-10994 99 9 , , PUNCT ajst-10994 99 10 3_2 3_2 NUM ajst-10994 99 11 , , PUNCT ajst-10994 99 12 3_4 3_4 NUM ajst-10994 99 13 , , PUNCT ajst-10994 99 14 and and CCONJ ajst-10994 99 15 2_3 2_3 NUM ajst-10994 99 16 from from ADP ajst-10994 99 17 the the DET ajst-10994 99 18 dataset dataset NOUN ajst-10994 99 19 for for ADP ajst-10994 99 20 analysis analysis NOUN ajst-10994 99 21 , , PUNCT ajst-10994 99 22 which which PRON ajst-10994 99 23 cover cover VERB ajst-10994 99 24 four four NUM ajst-10994 99 25 fault fault NOUN ajst-10994 99 26 modes mode NOUN ajst-10994 99 27 : : PUNCT ajst-10994 99 28 outer outer ADJ ajst-10994 99 29 race race NOUN ajst-10994 99 30 fault fault NOUN ajst-10994 99 31 ( ( PUNCT ajst-10994 99 32 or or CCONJ ajst-10994 99 33 ) ) PUNCT ajst-10994 99 34 , , PUNCT ajst-10994 99 35 inner inner ADJ ajst-10994 99 36 race race NOUN ajst-10994 99 37 fault fault NOUN ajst-10994 99 38 ( ( PUNCT ajst-10994 99 39 ir ir NOUN ajst-10994 99 40 ) ) PUNCT ajst-10994 99 41 , , PUNCT ajst-10994 99 42 cage cage NOUN ajst-10994 99 43 fault fault NOUN ajst-10994 99 44 , , PUNCT ajst-10994 99 45 and and CCONJ ajst-10994 99 46 combined combined ADJ ajst-10994 99 47 faults fault NOUN ajst-10994 99 48 ( ( PUNCT ajst-10994 99 49 ibco ibco PROPN ajst-10994 99 50 ) ) PUNCT ajst-10994 99 51 . . PUNCT ajst-10994 100 1 similar similar ADJ ajst-10994 100 2 to to ADP ajst-10994 100 3 the the DET ajst-10994 100 4 cwru cwru PROPN ajst-10994 100 5 dataset dataset PROPN ajst-10994 100 6 , , PUNCT ajst-10994 100 7 the the DET ajst-10994 100 8 original original ADJ ajst-10994 100 9 vibration vibration NOUN ajst-10994 100 10 signal signal NOUN ajst-10994 100 11 data datum NOUN ajst-10994 100 12 in in ADP ajst-10994 100 13 the the DET ajst-10994 100 14 xjtu xjtu PROPN ajst-10994 100 15 dataset dataset PROPN ajst-10994 100 16 has have VERB ajst-10994 100 17 a a DET ajst-10994 100 18 sampling sample VERB ajst-10994 100 19 frequency frequency NOUN ajst-10994 100 20 of of ADP ajst-10994 100 21 25.6khz 25.6khz PROPN ajst-10994 100 22 , , PUNCT ajst-10994 100 23 which which PRON ajst-10994 100 24 can can AUX ajst-10994 100 25 not not PART ajst-10994 100 26 be be AUX ajst-10994 100 27 directly directly ADV ajst-10994 100 28 used use VERB ajst-10994 100 29 as as ADP ajst-10994 100 30 input input NOUN ajst-10994 100 31 for for ADP ajst-10994 100 32 the the DET ajst-10994 100 33 tst tst NOUN ajst-10994 100 34 model model NOUN ajst-10994 100 35 . . PUNCT ajst-10994 101 1 therefore therefore ADV ajst-10994 101 2 , , PUNCT ajst-10994 101 3 it it PRON ajst-10994 101 4 also also ADV ajst-10994 101 5 needs need VERB ajst-10994 101 6 to to PART ajst-10994 101 7 undergo undergo VERB ajst-10994 101 8 preprocessing preprocessing NOUN ajst-10994 101 9 using use VERB ajst-10994 101 10 the the DET ajst-10994 101 11 resampling resample VERB ajst-10994 101 12 method method NOUN ajst-10994 101 13 depicted depict VERB ajst-10994 101 14 in in ADP ajst-10994 101 15 figure figure NOUN ajst-10994 101 16 3 3 NUM ajst-10994 101 17 . . PUNCT ajst-10994 102 1 after after ADP ajst-10994 102 2 preprocessing preprocesse VERB ajst-10994 102 3 , , PUNCT ajst-10994 102 4 the the DET ajst-10994 102 5 xjtu xjtu PROPN ajst-10994 102 6 dataset dataset PROPN ajst-10994 102 7 contains contain VERB ajst-10994 102 8 4000 4000 NUM ajst-10994 102 9 samples sample NOUN ajst-10994 102 10 , , PUNCT ajst-10994 102 11 with with ADP ajst-10994 102 12 2800 2800 NUM ajst-10994 102 13 samples sample NOUN ajst-10994 102 14 forming form VERB ajst-10994 102 15 the the DET ajst-10994 102 16 training training NOUN ajst-10994 102 17 set set NOUN ajst-10994 102 18 and and CCONJ ajst-10994 102 19 1200 1200 NUM ajst-10994 102 20 samples sample NOUN ajst-10994 102 21 forming form VERB ajst-10994 102 22 the the DET ajst-10994 102 23 test test NOUN ajst-10994 102 24 set set VERB ajst-10994 102 25 . . PUNCT ajst-10994 103 1 3.3 3.3 NUM ajst-10994 103 2 . . PUNCT ajst-10994 104 1 ucoon ucoon PROPN ajst-10994 104 2 data datum NOUN ajst-10994 104 3 set set VERB ajst-10994 104 4 the the DET ajst-10994 104 5 ucoon ucoon NOUN ajst-10994 104 6 dataset dataset NOUN ajst-10994 104 7 is be AUX ajst-10994 104 8 a a DET ajst-10994 104 9 gearbox gearbox NOUN ajst-10994 104 10 dataset dataset NOUN ajst-10994 104 11 collected collect VERB ajst-10994 104 12 and and CCONJ ajst-10994 104 13 organized organize VERB ajst-10994 104 14 by by ADP ajst-10994 104 15 the the DET ajst-10994 104 16 university university PROPN ajst-10994 104 17 of of ADP ajst-10994 104 18 connecticut connecticut PROPN ajst-10994 104 19 from from ADP ajst-10994 104 20 a a DET ajst-10994 104 21 two two NUM ajst-10994 104 22 - - PUNCT ajst-10994 104 23 stage stage NOUN ajst-10994 104 24 gear gear NOUN ajst-10994 104 25 transmission transmission NOUN ajst-10994 104 26 box box NOUN ajst-10994 104 27 . . PUNCT ajst-10994 105 1 the the DET ajst-10994 105 2 input input NOUN ajst-10994 105 3 shaft shaft NOUN ajst-10994 105 4 of of ADP ajst-10994 105 5 the the DET ajst-10994 105 6 gearbox gearbox NOUN ajst-10994 105 7 is be AUX ajst-10994 105 8 driven drive VERB ajst-10994 105 9 by by ADP ajst-10994 105 10 an an DET ajst-10994 105 11 ac ac PROPN ajst-10994 105 12 motor motor NOUN ajst-10994 105 13 , , PUNCT ajst-10994 105 14 and and CCONJ ajst-10994 105 15 the the DET ajst-10994 105 16 speed speed NOUN ajst-10994 105 17 of of ADP ajst-10994 105 18 the the DET ajst-10994 105 19 input input NOUN ajst-10994 105 20 shaft shaft NOUN ajst-10994 105 21 is be AUX ajst-10994 105 22 monitored monitor VERB ajst-10994 105 23 using use VERB ajst-10994 105 24 a a DET ajst-10994 105 25 tachometer tachometer NOUN ajst-10994 105 26 . . PUNCT ajst-10994 106 1 the the DET ajst-10994 106 2 first first ADJ ajst-10994 106 3 - - PUNCT ajst-10994 106 4 stage stage NOUN ajst-10994 106 5 input input NOUN ajst-10994 106 6 shaft shaft NOUN ajst-10994 106 7 is be AUX ajst-10994 106 8 equipped equip VERB ajst-10994 106 9 with with ADP ajst-10994 106 10 a a DET ajst-10994 106 11 32 32 NUM ajst-10994 106 12 - - PUNCT ajst-10994 106 13 tooth tooth NOUN ajst-10994 106 14 and and CCONJ ajst-10994 106 15 an an DET ajst-10994 106 16 80 80 NUM ajst-10994 106 17 - - PUNCT ajst-10994 106 18 tooth tooth NOUN ajst-10994 106 19 gear gear NOUN ajst-10994 106 20 , , PUNCT ajst-10994 106 21 while while SCONJ ajst-10994 106 22 the the DET ajst-10994 106 23 second second ADJ ajst-10994 106 24 - - PUNCT ajst-10994 106 25 stage stage NOUN ajst-10994 106 26 output output NOUN ajst-10994 106 27 shaft shaft NOUN ajst-10994 106 28 is be AUX ajst-10994 106 29 equipped equip VERB ajst-10994 106 30 with with ADP ajst-10994 106 31 a a DET ajst-10994 106 32 48 48 NUM ajst-10994 106 33 - - PUNCT ajst-10994 106 34 tooth tooth NOUN ajst-10994 106 35 and and CCONJ ajst-10994 106 36 a a DET ajst-10994 106 37 64 64 NUM ajst-10994 106 38 - - PUNCT ajst-10994 106 39 tooth tooth NOUN ajst-10994 106 40 gear[9 gear[9 NOUN ajst-10994 106 41 ] ] PUNCT ajst-10994 106 42 . . PUNCT ajst-10994 107 1 the the DET ajst-10994 107 2 quiver quiver PROPN ajst-10994 107 3 signal signal PROPN ajst-10994 107 4 data datum NOUN ajst-10994 107 5 of of ADP ajst-10994 107 6 the the DET ajst-10994 107 7 entire entire ADJ ajst-10994 107 8 gearbox gearbox NOUN ajst-10994 107 9 is be AUX ajst-10994 107 10 collected collect VERB ajst-10994 107 11 using use VERB ajst-10994 107 12 an an DET ajst-10994 107 13 measuring measure VERB ajst-10994 107 14 device device NOUN ajst-10994 107 15 fixed fix VERB ajst-10994 107 16 on on ADP ajst-10994 107 17 the the DET ajst-10994 107 18 gearbox gearbox NOUN ajst-10994 107 19 housing housing NOUN ajst-10994 107 20 , , PUNCT ajst-10994 107 21 with with ADP ajst-10994 107 22 a a DET ajst-10994 107 23 sampling sample VERB ajst-10994 107 24 frequency frequency NOUN ajst-10994 107 25 of of ADP ajst-10994 107 26 20khz 20khz PROPN ajst-10994 107 27 . . PUNCT ajst-10994 108 1 each each DET ajst-10994 108 2 sample sample NOUN ajst-10994 108 3 in in ADP ajst-10994 108 4 the the DET ajst-10994 108 5 dataset dataset NOUN ajst-10994 108 6 has have VERB ajst-10994 108 7 a a DET ajst-10994 108 8 sampling sampling NOUN ajst-10994 108 9 length length NOUN ajst-10994 108 10 of of ADP ajst-10994 108 11 3600 3600 NUM ajst-10994 108 12 data datum NOUN ajst-10994 108 13 points point NOUN ajst-10994 108 14 . . PUNCT ajst-10994 109 1 to to PART ajst-10994 109 2 fit fit VERB ajst-10994 109 3 the the DET ajst-10994 109 4 input input NOUN ajst-10994 109 5 dimension dimension NOUN ajst-10994 109 6 of of ADP ajst-10994 109 7 the the DET ajst-10994 109 8 tst tst NOUN ajst-10994 109 9 model model NOUN ajst-10994 109 10 , , PUNCT ajst-10994 109 11 we we PRON ajst-10994 109 12 selected select VERB ajst-10994 109 13 the the DET ajst-10994 109 14 middle middle ADJ ajst-10994 109 15 2048 2048 NUM ajst-10994 109 16 data datum NOUN ajst-10994 109 17 points point NOUN ajst-10994 109 18 of of ADP ajst-10994 109 19 each each DET ajst-10994 109 20 sample sample NOUN ajst-10994 109 21 as as ADP ajst-10994 109 22 the the DET ajst-10994 109 23 data datum NOUN ajst-10994 109 24 for for ADP ajst-10994 109 25 analysis analysis NOUN ajst-10994 109 26 in in ADP ajst-10994 109 27 this this DET ajst-10994 109 28 chapter chapter NOUN ajst-10994 109 29 . . PUNCT ajst-10994 110 1 the the DET ajst-10994 110 2 ucoon ucoon PROPN ajst-10994 110 3 dataset dataset VERB ajst-10994 110 4 consists consist NOUN ajst-10994 110 5 of of ADP ajst-10994 110 6 936 936 NUM ajst-10994 110 7 samples sample NOUN ajst-10994 110 8 and and CCONJ ajst-10994 110 9 contains contain VERB ajst-10994 110 10 9 9 NUM ajst-10994 110 11 fault fault NOUN ajst-10994 110 12 modes mode NOUN ajst-10994 110 13 , , PUNCT ajst-10994 110 14 namely namely ADV ajst-10994 110 15 normal normal ADJ ajst-10994 110 16 condition condition NOUN ajst-10994 110 17 ( ( PUNCT ajst-10994 110 18 nc nc PROPN ajst-10994 110 19 ) ) PUNCT ajst-10994 110 20 , , PUNCT ajst-10994 110 21 missing miss VERB ajst-10994 110 22 tooth tooth NOUN ajst-10994 110 23 , , PUNCT ajst-10994 110 24 crack crack NOUN ajst-10994 110 25 , , PUNCT ajst-10994 110 26 spall spall NOUN ajst-10994 110 27 , , PUNCT ajst-10994 110 28 and and CCONJ ajst-10994 110 29 5 5 NUM ajst-10994 110 30 different different ADJ ajst-10994 110 31 degrees degree NOUN ajst-10994 110 32 of of ADP ajst-10994 110 33 chipped chip VERB ajst-10994 110 34 teeth tooth NOUN ajst-10994 110 35 ( ( PUNCT ajst-10994 110 36 referred refer VERB ajst-10994 110 37 to to ADP ajst-10994 110 38 as as ADP ajst-10994 110 39 chip5a chip5a NOUN ajst-10994 110 40 to to ADP ajst-10994 110 41 chip1a chip1a VERB ajst-10994 110 42 ) ) PUNCT ajst-10994 110 43 . . PUNCT ajst-10994 111 1 for for ADP ajst-10994 111 2 this this DET ajst-10994 111 3 paper paper NOUN ajst-10994 111 4 , , PUNCT ajst-10994 111 5 we we PRON ajst-10994 111 6 randomly randomly ADV ajst-10994 111 7 selected select VERB ajst-10994 111 8 655 655 NUM ajst-10994 111 9 samples sample NOUN ajst-10994 111 10 as as ADP ajst-10994 111 11 the the DET ajst-10994 111 12 training training NOUN ajst-10994 111 13 set set NOUN ajst-10994 111 14 and and CCONJ ajst-10994 111 15 the the DET ajst-10994 111 16 remaining remain VERB ajst-10994 111 17 281 281 NUM ajst-10994 111 18 samples sample NOUN ajst-10994 111 19 as as ADP ajst-10994 111 20 the the DET ajst-10994 111 21 test test NOUN ajst-10994 111 22 set[10 set[10 NUM ajst-10994 111 23 ] ] PUNCT ajst-10994 111 24 . . PUNCT ajst-10994 112 1 4 4 X ajst-10994 112 2 . . X ajst-10994 112 3 simulation simulation NOUN ajst-10994 112 4 analysis analysis NOUN ajst-10994 112 5 4.1 4.1 NUM ajst-10994 112 6 . . PUNCT ajst-10994 113 1 fault fault VERB ajst-10994 113 2 diagnosis diagnosis NOUN ajst-10994 113 3 results result NOUN ajst-10994 113 4 of of ADP ajst-10994 113 5 tst tst NOUN ajst-10994 113 6 model model NOUN ajst-10994 113 7 on on ADP ajst-10994 113 8 a a DET ajst-10994 113 9 given give VERB ajst-10994 113 10 data datum NOUN ajst-10994 113 11 set set VERB ajst-10994 113 12 this this DET ajst-10994 113 13 article article NOUN ajst-10994 113 14 will will AUX ajst-10994 113 15 provide provide VERB ajst-10994 113 16 a a DET ajst-10994 113 17 detailed detailed ADJ ajst-10994 113 18 discussion discussion NOUN ajst-10994 113 19 and and CCONJ ajst-10994 113 20 analysis analysis NOUN ajst-10994 113 21 of of ADP ajst-10994 113 22 the the DET ajst-10994 113 23 fault fault NOUN ajst-10994 113 24 diagnosis diagnosis NOUN ajst-10994 113 25 results result NOUN ajst-10994 113 26 of of ADP ajst-10994 113 27 the the DET ajst-10994 113 28 proposed propose VERB ajst-10994 113 29 tst tst NOUN ajst-10994 113 30 model model NOUN ajst-10994 113 31 , , PUNCT ajst-10994 113 32 including include VERB ajst-10994 113 33 the the DET ajst-10994 113 34 basic basic ADJ ajst-10994 113 35 fault fault NOUN ajst-10994 113 36 diagnosis diagnosis NOUN ajst-10994 113 37 outcomes outcome NOUN ajst-10994 113 38 and and CCONJ ajst-10994 113 39 a a DET ajst-10994 113 40 relative relative NOUN ajst-10994 113 41 to to ADP ajst-10994 113 42 unrelated unrelated ADJ ajst-10994 113 43 existing exist VERB ajst-10994 113 44 fault fault NOUN ajst-10994 113 45 diagnosis diagnosis NOUN ajst-10994 113 46 methods[11 methods[11 NOUN ajst-10994 113 47 ] ] PUNCT ajst-10994 113 48 . . PUNCT ajst-10994 114 1 the the DET ajst-10994 114 2 hardware hardware NOUN ajst-10994 114 3 environment environment NOUN ajst-10994 114 4 used use VERB ajst-10994 114 5 to to PART ajst-10994 114 6 run run VERB ajst-10994 114 7 the the DET ajst-10994 114 8 code code NOUN ajst-10994 114 9 includes include VERB ajst-10994 114 10 an an DET ajst-10994 114 11 amd amd ADJ ajst-10994 114 12 ryzen ryzen ADJ ajst-10994 114 13 threadripperpro threadripperpro VERB ajst-10994 114 14 3995wx 3995wx ADJ ajst-10994 114 15 cpu cpu NOUN ajst-10994 114 16 and and CCONJ ajst-10994 114 17 an an DET ajst-10994 114 18 nvidia nvidia PROPN ajst-10994 114 19 rtx rtx PROPN ajst-10994 114 20 3090 3090 NUM ajst-10994 114 21 gpu gpu PROPN ajst-10994 114 22 . . PUNCT ajst-10994 115 1 the the DET ajst-10994 115 2 deep deep ADJ ajst-10994 115 3 learning learning NOUN ajst-10994 115 4 framework framework NOUN ajst-10994 115 5 and and CCONJ ajst-10994 115 6 software software NOUN ajst-10994 115 7 environment environment NOUN ajst-10994 115 8 utilized utilize VERB ajst-10994 115 9 are be AUX ajst-10994 115 10 python python NOUN ajst-10994 115 11 3.8 3.8 NUM ajst-10994 115 12 , , PUNCT ajst-10994 115 13 pytorch pytorch NOUN ajst-10994 115 14 1.8.1 1.8.1 NUM ajst-10994 115 15 , , PUNCT ajst-10994 115 16 and and CCONJ ajst-10994 115 17 cuda cuda PROPN ajst-10994 115 18 10.2 10.2 NUM ajst-10994 115 19 . . PUNCT ajst-10994 116 1 during during ADP ajst-10994 116 2 the the DET ajst-10994 116 3 training training NOUN ajst-10994 116 4 process process NOUN ajst-10994 116 5 , , PUNCT ajst-10994 116 6 considering consider VERB ajst-10994 116 7 that that SCONJ ajst-10994 116 8 the the DET ajst-10994 116 9 cwru cwru PROPN ajst-10994 116 10 , , PUNCT ajst-10994 116 11 xjtu xjtu PROPN ajst-10994 116 12 , , PUNCT ajst-10994 116 13 and and CCONJ ajst-10994 116 14 ucoon ucoon NOUN ajst-10994 116 15 datasets dataset NOUN ajst-10994 116 16 have have VERB ajst-10994 116 17 different different ADJ ajst-10994 116 18 scales scale NOUN ajst-10994 116 19 , , PUNCT ajst-10994 116 20 the the DET ajst-10994 116 21 batch batch NOUN ajst-10994 116 22 sizes size NOUN ajst-10994 116 23 for for ADP ajst-10994 116 24 the the DET ajst-10994 116 25 three three NUM ajst-10994 116 26 datasets dataset NOUN ajst-10994 116 27 were be AUX ajst-10994 116 28 set set VERB ajst-10994 116 29 to to ADP ajst-10994 116 30 128 128 NUM ajst-10994 116 31 , , PUNCT ajst-10994 116 32 64 64 NUM ajst-10994 116 33 , , PUNCT ajst-10994 116 34 and and CCONJ ajst-10994 116 35 32 32 NUM ajst-10994 116 36 , , PUNCT ajst-10994 116 37 respectively[12 respectively[12 NOUN ajst-10994 116 38 ] ] PUNCT ajst-10994 116 39 . . PUNCT ajst-10994 117 1 to to PART ajst-10994 117 2 make make VERB ajst-10994 117 3 the the DET ajst-10994 117 4 training training NOUN ajst-10994 117 5 process process NOUN ajst-10994 117 6 more more ADV ajst-10994 117 7 generalized generalize VERB ajst-10994 117 8 , , PUNCT ajst-10994 117 9 no no DET ajst-10994 117 10 model model NOUN ajst-10994 117 11 selection selection NOUN ajst-10994 117 12 strategies strategy NOUN ajst-10994 117 13 like like ADP ajst-10994 117 14 early early ADJ ajst-10994 117 15 stopping stopping NOUN ajst-10994 117 16 were be AUX ajst-10994 117 17 employed employ VERB ajst-10994 117 18 , , PUNCT ajst-10994 117 19 and and CCONJ ajst-10994 117 20 the the DET ajst-10994 117 21 tst tst NOUN ajst-10994 117 22 model model NOUN ajst-10994 117 23 was be AUX ajst-10994 117 24 trained train VERB ajst-10994 117 25 for for ADP ajst-10994 117 26 50 50 NUM ajst-10994 117 27 epochs epoch NOUN ajst-10994 117 28 without without ADP ajst-10994 117 29 early early ADJ ajst-10994 117 30 termination termination NOUN ajst-10994 117 31 . . PUNCT ajst-10994 118 1 furthermore furthermore ADV ajst-10994 118 2 , , PUNCT ajst-10994 118 3 to to PART ajst-10994 118 4 mitigate mitigate VERB ajst-10994 118 5 the the DET ajst-10994 118 6 consequences consequence NOUN ajst-10994 118 7 of of ADP ajst-10994 118 8 random random ADJ ajst-10994 118 9 starting starting NOUN ajst-10994 118 10 values value NOUN ajst-10994 118 11 , , PUNCT ajst-10994 118 12 the the DET ajst-10994 118 13 tst tst NOUN ajst-10994 118 14 model model NOUN ajst-10994 118 15 was be AUX ajst-10994 118 16 repetitively repetitively ADV ajst-10994 118 17 trained train VERB ajst-10994 118 18 100 100 NUM ajst-10994 118 19 times time NOUN ajst-10994 118 20 on on ADP ajst-10994 118 21 each each DET ajst-10994 118 22 dataset dataset NOUN ajst-10994 118 23 with with ADP ajst-10994 118 24 the the DET ajst-10994 118 25 same same ADJ ajst-10994 118 26 parameter parameter NOUN ajst-10994 118 27 settings setting NOUN ajst-10994 118 28 . . PUNCT ajst-10994 119 1 the the DET ajst-10994 119 2 results result NOUN ajst-10994 119 3 of of ADP ajst-10994 119 4 the the DET ajst-10994 119 5 loss loss NOUN ajst-10994 119 6 function function NOUN ajst-10994 119 7 and and CCONJ ajst-10994 119 8 fault fault VERB ajst-10994 119 9 diagnosis diagnosis NOUN ajst-10994 119 10 accuracy accuracy NOUN ajst-10994 119 11 are be AUX ajst-10994 119 12 presented present VERB ajst-10994 119 13 in in ADP ajst-10994 119 14 the the DET ajst-10994 119 15 form form NOUN ajst-10994 119 16 of of ADP ajst-10994 119 17 both both CCONJ ajst-10994 119 18 average average ADJ ajst-10994 119 19 values value NOUN ajst-10994 119 20 and and CCONJ ajst-10994 119 21 box box NOUN ajst-10994 119 22 plots plot NOUN ajst-10994 119 23 with with ADP ajst-10994 119 24 statistical statistical ADJ ajst-10994 119 25 information information NOUN ajst-10994 119 26 . . PUNCT ajst-10994 120 1 figures figure VERB ajst-10994 120 2 4 4 NUM ajst-10994 120 3 to to PART ajst-10994 120 4 9 9 NUM ajst-10994 120 5 illustrate illustrate VERB ajst-10994 120 6 the the DET ajst-10994 120 7 training training NOUN ajst-10994 120 8 process process NOUN ajst-10994 120 9 of of ADP ajst-10994 120 10 the the DET ajst-10994 120 11 tst tst NOUN ajst-10994 120 12 model model NOUN ajst-10994 120 13 on on ADP ajst-10994 120 14 the the DET ajst-10994 120 15 three three NUM ajst-10994 120 16 datasets dataset NOUN ajst-10994 120 17 , , PUNCT ajst-10994 120 18 including include VERB ajst-10994 120 19 the the DET ajst-10994 120 20 variations variation NOUN ajst-10994 120 21 of of ADP ajst-10994 120 22 the the DET ajst-10994 120 23 loss loss NOUN ajst-10994 120 24 function function NOUN ajst-10994 120 25 and and CCONJ ajst-10994 120 26 recognition recognition NOUN ajst-10994 120 27 accuracy accuracy NOUN ajst-10994 120 28 on on ADP ajst-10994 120 29 both both CCONJ ajst-10994 120 30 the the DET ajst-10994 120 31 training training NOUN ajst-10994 120 32 and and CCONJ ajst-10994 120 33 testing testing NOUN ajst-10994 120 34 sets set NOUN ajst-10994 120 35 . . PUNCT ajst-10994 121 1 in in ADP ajst-10994 121 2 the the DET ajst-10994 121 3 early early ADJ ajst-10994 121 4 stages stage NOUN ajst-10994 121 5 of of ADP ajst-10994 121 6 training training NOUN ajst-10994 121 7 , , PUNCT ajst-10994 121 8 the the DET ajst-10994 121 9 impact impact NOUN ajst-10994 121 10 of of ADP ajst-10994 121 11 random random ADJ ajst-10994 121 12 initialization initialization NOUN ajst-10994 121 13 could could AUX ajst-10994 121 14 be be AUX ajst-10994 121 15 significant significant ADJ ajst-10994 121 16 , , PUNCT ajst-10994 121 17 which which PRON ajst-10994 121 18 is be AUX ajst-10994 121 19 why why SCONJ ajst-10994 121 20 the the DET ajst-10994 121 21 changes change NOUN ajst-10994 121 22 in in ADP ajst-10994 121 23 the the DET ajst-10994 121 24 loss loss NOUN ajst-10994 121 25 function function NOUN ajst-10994 121 26 and and CCONJ ajst-10994 121 27 accuracy accuracy NOUN ajst-10994 121 28 are be AUX ajst-10994 121 29 displayed display VERB ajst-10994 121 30 using use VERB ajst-10994 121 31 box box PROPN ajst-10994 121 32 plots[12 plots[12 PROPN ajst-10994 121 33 ] ] PUNCT ajst-10994 121 34 . . PUNCT ajst-10994 122 1 however however ADV ajst-10994 122 2 , , PUNCT ajst-10994 122 3 the the DET ajst-10994 122 4 overall overall ADJ ajst-10994 122 5 variations variation NOUN ajst-10994 122 6 of of ADP ajst-10994 122 7 the the DET ajst-10994 122 8 loss loss NOUN ajst-10994 122 9 function function NOUN ajst-10994 122 10 and and CCONJ ajst-10994 122 11 accuracy accuracy NOUN ajst-10994 122 12 throughout throughout ADP ajst-10994 122 13 the the DET ajst-10994 122 14 entire entire ADJ ajst-10994 122 15 training training NOUN ajst-10994 122 16 process process NOUN ajst-10994 122 17 are be AUX ajst-10994 122 18 represented represent VERB ajst-10994 122 19 using use VERB ajst-10994 122 20 the the DET ajst-10994 122 21 average average ADJ ajst-10994 122 22 values value NOUN ajst-10994 122 23 of of ADP ajst-10994 122 24 the the DET ajst-10994 122 25 repeated repeat VERB ajst-10994 122 26 training training NOUN ajst-10994 122 27 results result NOUN ajst-10994 122 28 . . PUNCT ajst-10994 123 1 from from ADP ajst-10994 123 2 the the DET ajst-10994 123 3 results result NOUN ajst-10994 123 4 in in ADP ajst-10994 123 5 figure figure NOUN ajst-10994 123 6 4(a 4(a NUM ajst-10994 123 7 ) ) PUNCT ajst-10994 123 8 and and CCONJ ajst-10994 123 9 figure figure VERB ajst-10994 123 10 4(b 4(b NUM ajst-10994 123 11 ) ) PUNCT ajst-10994 123 12 on on ADP ajst-10994 123 13 the the DET ajst-10994 123 14 cwru cwru PROPN ajst-10994 123 15 dataset dataset NOUN ajst-10994 123 16 , , PUNCT ajst-10994 123 17 it it PRON ajst-10994 123 18 can can AUX ajst-10994 123 19 be be AUX ajst-10994 123 20 observed observe VERB ajst-10994 123 21 that that SCONJ ajst-10994 123 22 different different ADJ ajst-10994 123 23 random random ADJ ajst-10994 123 24 initializations initialization NOUN ajst-10994 123 25 of of ADP ajst-10994 123 26 parameters parameter NOUN ajst-10994 123 27 lead lead VERB ajst-10994 123 28 to to ADP ajst-10994 123 29 significant significant ADJ ajst-10994 123 30 fluctuations fluctuation NOUN ajst-10994 123 31 in in ADP ajst-10994 123 32 the the DET ajst-10994 123 33 model model NOUN ajst-10994 123 34 's 's PART ajst-10994 123 35 loss loss NOUN ajst-10994 123 36 function function NOUN ajst-10994 123 37 during during ADP ajst-10994 123 38 the the DET ajst-10994 123 39 early early ADJ ajst-10994 123 40 stages stage NOUN ajst-10994 123 41 of of ADP ajst-10994 123 42 training training NOUN ajst-10994 123 43 . . PUNCT ajst-10994 124 1 however however ADV ajst-10994 124 2 , , PUNCT ajst-10994 124 3 after after ADP ajst-10994 124 4 approximately approximately ADV ajst-10994 124 5 10 10 NUM ajst-10994 124 6 epochs epoch NOUN ajst-10994 124 7 of of ADP ajst-10994 124 8 training training NOUN ajst-10994 124 9 , , PUNCT ajst-10994 124 10 the the DET ajst-10994 124 11 box box NOUN ajst-10994 124 12 plots plot NOUN ajst-10994 124 13 of of ADP ajst-10994 124 14 the the DET ajst-10994 124 15 loss loss NOUN ajst-10994 124 16 function function NOUN ajst-10994 124 17 values value NOUN ajst-10994 124 18 on on ADP ajst-10994 124 19 both both CCONJ ajst-10994 124 20 the the DET ajst-10994 124 21 training training NOUN ajst-10994 124 22 and and CCONJ ajst-10994 124 23 testing testing NOUN ajst-10994 124 24 sets set NOUN ajst-10994 124 25 become become VERB ajst-10994 124 26 more more ADV ajst-10994 124 27 flattened flatten VERB ajst-10994 124 28 , , PUNCT ajst-10994 124 29 and and CCONJ ajst-10994 124 30 the the DET ajst-10994 124 31 number number NOUN ajst-10994 124 32 of of ADP ajst-10994 124 33 outliers outlier NOUN ajst-10994 124 34 ( ( PUNCT ajst-10994 124 35 depicted depict VERB ajst-10994 124 36 as as ADP ajst-10994 124 37 black black ADJ ajst-10994 124 38 dots dot NOUN ajst-10994 124 39 in in ADP ajst-10994 124 40 figure figure NOUN ajst-10994 124 41 4(a 4(a NUM ajst-10994 124 42 ) ) PUNCT ajst-10994 124 43 and and CCONJ ajst-10994 124 44 figure figure VERB ajst-10994 124 45 4(b 4(b NUM ajst-10994 124 46 ) ) PUNCT ajst-10994 124 47 ) ) PUNCT ajst-10994 124 48 noticeably noticeably ADV ajst-10994 124 49 decreases decrease VERB ajst-10994 124 50 . . PUNCT ajst-10994 125 1 this this PRON ajst-10994 125 2 indicates indicate VERB ajst-10994 125 3 that that SCONJ ajst-10994 125 4 the the DET ajst-10994 125 5 impact impact NOUN ajst-10994 125 6 of of ADP ajst-10994 125 7 random random ADJ ajst-10994 125 8 initialization initialization NOUN ajst-10994 125 9 has have AUX ajst-10994 125 10 been be AUX ajst-10994 125 11 largely largely ADV ajst-10994 125 12 mitigated mitigate VERB ajst-10994 125 13 , , PUNCT ajst-10994 125 14 and and CCONJ ajst-10994 125 15 the the DET ajst-10994 125 16 loss loss NOUN ajst-10994 125 17 function function NOUN ajst-10994 125 18 on on ADP ajst-10994 125 19 the the DET ajst-10994 125 20 training training NOUN ajst-10994 125 21 and and CCONJ ajst-10994 125 22 testing testing NOUN ajst-10994 125 23 sets set NOUN ajst-10994 125 24 has have AUX ajst-10994 125 25 gradually gradually ADV ajst-10994 125 26 stabilized stabilize VERB ajst-10994 125 27 . . PUNCT ajst-10994 126 1 these these DET ajst-10994 126 2 results result NOUN ajst-10994 126 3 demonstrate demonstrate VERB ajst-10994 126 4 that that SCONJ ajst-10994 126 5 the the DET ajst-10994 126 6 tst tst NOUN ajst-10994 126 7 model model NOUN ajst-10994 126 8 exhibits exhibit VERB ajst-10994 126 9 good good ADJ ajst-10994 126 10 robustness robustness NOUN ajst-10994 126 11 in in ADP ajst-10994 126 12 fault fault NOUN ajst-10994 126 13 diagnosis diagnosis NOUN ajst-10994 126 14 performance performance NOUN ajst-10994 126 15 , , PUNCT ajst-10994 126 16 as as SCONJ ajst-10994 126 17 it it PRON ajst-10994 126 18 can can AUX ajst-10994 126 19 effectively effectively ADV ajst-10994 126 20 mitigate mitigate VERB ajst-10994 126 21 the the DET ajst-10994 126 22 effects effect NOUN ajst-10994 126 23 of of ADP ajst-10994 126 24 random random ADJ ajst-10994 126 25 initialization initialization NOUN ajst-10994 126 26 to to ADP ajst-10994 126 27 a a DET ajst-10994 126 28 certain certain ADJ ajst-10994 126 29 extent extent NOUN ajst-10994 126 30 . . PUNCT ajst-10994 127 1 in in ADP ajst-10994 127 2 addition addition NOUN ajst-10994 127 3 , , PUNCT ajst-10994 127 4 the the DET ajst-10994 127 5 variation variation NOUN ajst-10994 127 6 of of ADP ajst-10994 127 7 the the DET ajst-10994 127 8 average average ADJ ajst-10994 127 9 loss loss NOUN ajst-10994 127 10 function function NOUN ajst-10994 127 11 value value NOUN ajst-10994 127 12 ( ( PUNCT ajst-10994 127 13 denoted denote VERB ajst-10994 127 14 as as ADP ajst-10994 127 15 avgloss avgloss NOUN ajst-10994 127 16 ) ) PUNCT ajst-10994 127 17 throughout throughout ADP ajst-10994 127 18 the the DET ajst-10994 127 19 entire entire ADJ ajst-10994 127 20 training training NOUN ajst-10994 127 21 process process NOUN ajst-10994 127 22 is be AUX ajst-10994 127 23 shown show VERB ajst-10994 127 24 in in ADP ajst-10994 127 25 figure figure NOUN ajst-10994 127 26 4(c 4(c NUM ajst-10994 127 27 ) ) PUNCT ajst-10994 127 28 . . PUNCT ajst-10994 128 1 it it PRON ajst-10994 128 2 can can AUX ajst-10994 128 3 be be AUX ajst-10994 128 4 observed observe VERB ajst-10994 128 5 that that SCONJ ajst-10994 128 6 the the DET ajst-10994 128 7 tst tst NOUN ajst-10994 128 8 model model NOUN ajst-10994 128 9 demonstrates demonstrate VERB ajst-10994 128 10 a a DET ajst-10994 128 11 decaying decay VERB ajst-10994 128 12 trend trend NOUN ajst-10994 128 13 of of ADP ajst-10994 128 14 the the DET ajst-10994 128 15 loss loss NOUN ajst-10994 128 16 function function NOUN ajst-10994 128 17 values value NOUN ajst-10994 128 18 on on ADP ajst-10994 128 19 both both CCONJ ajst-10994 128 20 the the DET ajst-10994 128 21 training training NOUN ajst-10994 128 22 and and CCONJ ajst-10994 128 23 testing testing NOUN ajst-10994 128 24 sets set NOUN ajst-10994 128 25 , , PUNCT ajst-10994 128 26 and and CCONJ ajst-10994 128 27 there there PRON ajst-10994 128 28 are be VERB ajst-10994 128 29 no no DET ajst-10994 128 30 significant significant ADJ ajst-10994 128 31 wide wide ADJ ajst-10994 128 32 fluctuations fluctuation NOUN ajst-10994 128 33 . . PUNCT ajst-10994 129 1 this this PRON ajst-10994 129 2 suggests suggest VERB ajst-10994 129 3 that that SCONJ ajst-10994 129 4 the the DET ajst-10994 129 5 choice choice NOUN ajst-10994 129 6 of of ADP ajst-10994 129 7 optimizer optimizer NOUN ajst-10994 129 8 parameters parameter NOUN ajst-10994 129 9 and and CCONJ ajst-10994 129 10 the the DET ajst-10994 129 11 design design NOUN ajst-10994 129 12 of of ADP ajst-10994 129 13 the the DET ajst-10994 129 14 model model NOUN ajst-10994 129 15 architecture architecture NOUN ajst-10994 129 16 are be AUX ajst-10994 129 17 reasonably reasonably ADV ajst-10994 129 18 appropriate appropriate ADJ ajst-10994 129 19 , , PUNCT ajst-10994 129 20 resulting result VERB ajst-10994 129 21 in in ADP ajst-10994 129 22 a a DET ajst-10994 129 23 reliable reliable ADJ ajst-10994 129 24 gradient gradient NOUN ajst-10994 129 25 optimization optimization NOUN ajst-10994 129 26 during during ADP ajst-10994 129 27 the the DET ajst-10994 129 28 training training NOUN ajst-10994 129 29 process process NOUN ajst-10994 129 30 . . PUNCT ajst-10994 130 1 when when SCONJ ajst-10994 130 2 training training NOUN ajst-10994 130 3 concludes conclude VERB ajst-10994 130 4 , , PUNCT ajst-10994 130 5 the the DET ajst-10994 130 6 average average ADJ ajst-10994 130 7 loss loss NOUN ajst-10994 130 8 function function NOUN ajst-10994 130 9 values value NOUN ajst-10994 130 10 on on ADP ajst-10994 130 11 the the DET ajst-10994 130 12 training training NOUN ajst-10994 130 13 and and CCONJ ajst-10994 130 14 testing testing NOUN ajst-10994 130 15 sets set NOUN ajst-10994 130 16 are be AUX ajst-10994 130 17 relatively relatively ADV ajst-10994 130 18 close close ADJ ajst-10994 130 19 , , PUNCT ajst-10994 130 20 indicating indicate VERB ajst-10994 130 21 that that SCONJ ajst-10994 130 22 the the DET ajst-10994 130 23 model model NOUN ajst-10994 130 24 did do AUX ajst-10994 130 25 not not PART ajst-10994 130 26 suffer suffer VERB ajst-10994 130 27 from from ADP ajst-10994 130 28 severe severe ADJ ajst-10994 130 29 overfitting overfitting NOUN ajst-10994 130 30 . . PUNCT ajst-10994 131 1 overall overall ADV ajst-10994 131 2 , , PUNCT ajst-10994 131 3 these these DET ajst-10994 131 4 findings finding NOUN ajst-10994 131 5 demonstrate demonstrate VERB ajst-10994 131 6 that that SCONJ ajst-10994 131 7 the the DET ajst-10994 131 8 tst tst NOUN ajst-10994 131 9 model model NOUN ajst-10994 131 10 exhibits exhibit VERB ajst-10994 131 11 a a DET ajst-10994 131 12 good good ADJ ajst-10994 131 13 level level NOUN ajst-10994 131 14 of of ADP ajst-10994 131 15 robustness robustness NOUN ajst-10994 131 16 and and CCONJ ajst-10994 131 17 generalization generalization NOUN ajst-10994 131 18 in in ADP ajst-10994 131 19 fault fault NOUN ajst-10994 131 20 diagnosis diagnosis NOUN ajst-10994 131 21 performance performance NOUN ajst-10994 131 22 , , PUNCT ajst-10994 131 23 effectively effectively ADV ajst-10994 131 24 dealing deal VERB ajst-10994 131 25 with with ADP ajst-10994 131 26 the the DET ajst-10994 131 27 challenges challenge NOUN ajst-10994 131 28 posed pose VERB ajst-10994 131 29 by by ADP ajst-10994 131 30 random random ADJ ajst-10994 131 31 initialization initialization NOUN ajst-10994 131 32 and and CCONJ ajst-10994 131 33 achieving achieve VERB ajst-10994 131 34 a a DET ajst-10994 131 35 balanced balanced ADJ ajst-10994 131 36 performance performance NOUN ajst-10994 131 37 between between ADP ajst-10994 131 38 the the DET ajst-10994 131 39 training training NOUN ajst-10994 131 40 and and CCONJ ajst-10994 131 41 testing testing NOUN ajst-10994 131 42 sets set NOUN ajst-10994 131 43 . . PUNCT ajst-10994 132 1 83 83 NUM ajst-10994 132 2 figure figure NOUN ajst-10994 132 3 4 4 NUM ajst-10994 132 4 . . PUNCT ajst-10994 132 5 changes change NOUN ajst-10994 132 6 in in ADP ajst-10994 132 7 the the DET ajst-10994 132 8 loss loss NOUN ajst-10994 132 9 function function NOUN ajst-10994 132 10 on on ADP ajst-10994 132 11 the the DET ajst-10994 132 12 cwru cwru PROPN ajst-10994 132 13 data datum NOUN ajst-10994 132 14 set set VERB ajst-10994 132 15 during during ADP ajst-10994 132 16 the the DET ajst-10994 132 17 training training NOUN ajst-10994 132 18 process process NOUN ajst-10994 132 19 ( ( PUNCT ajst-10994 132 20 a a X ajst-10994 132 21 ) ) PUNCT ajst-10994 132 22 box box NOUN ajst-10994 132 23 diagram diagram NOUN ajst-10994 132 24 of of ADP ajst-10994 132 25 the the DET ajst-10994 132 26 first first ADJ ajst-10994 132 27 10 10 NUM ajst-10994 132 28 generations generation NOUN ajst-10994 132 29 of of ADP ajst-10994 132 30 the the DET ajst-10994 132 31 loss loss NOUN ajst-10994 132 32 function function NOUN ajst-10994 132 33 in in ADP ajst-10994 132 34 the the DET ajst-10994 132 35 training training NOUN ajst-10994 132 36 set set NOUN ajst-10994 132 37 ( ( PUNCT ajst-10994 132 38 b b NOUN ajst-10994 132 39 ) ) PUNCT ajst-10994 132 40 box box NOUN ajst-10994 132 41 diagram diagram NOUN ajst-10994 132 42 of of ADP ajst-10994 132 43 the the DET ajst-10994 132 44 first first ADJ ajst-10994 132 45 10 10 NUM ajst-10994 132 46 generations generation NOUN ajst-10994 132 47 of of ADP ajst-10994 132 48 the the DET ajst-10994 132 49 loss loss NOUN ajst-10994 132 50 function function NOUN ajst-10994 132 51 in in ADP ajst-10994 132 52 the the DET ajst-10994 132 53 test test NOUN ajst-10994 132 54 set set NOUN ajst-10994 132 55 ( ( PUNCT ajst-10994 132 56 c c X ajst-10994 132 57 ) ) PUNCT ajst-10994 132 58 the the DET ajst-10994 132 59 average average ADJ ajst-10994 132 60 loss loss NOUN ajst-10994 132 61 function function NOUN ajst-10994 132 62 value value NOUN ajst-10994 132 63 during during ADP ajst-10994 132 64 the the DET ajst-10994 132 65 entire entire ADJ ajst-10994 132 66 training training NOUN ajst-10994 132 67 process process NOUN ajst-10994 132 68 similarly similarly ADV ajst-10994 132 69 , , PUNCT ajst-10994 132 70 in in ADP ajst-10994 132 71 figure figure NOUN ajst-10994 132 72 5 5 NUM ajst-10994 132 73 and and CCONJ ajst-10994 132 74 figure figure VERB ajst-10994 132 75 6 6 NUM ajst-10994 132 76 , , PUNCT ajst-10994 132 77 similar similar ADJ ajst-10994 132 78 phenomena phenomenon NOUN ajst-10994 132 79 can can AUX ajst-10994 132 80 be be AUX ajst-10994 132 81 observed observe VERB ajst-10994 132 82 . . PUNCT ajst-10994 133 1 however however ADV ajst-10994 133 2 , , PUNCT ajst-10994 133 3 it it PRON ajst-10994 133 4 should should AUX ajst-10994 133 5 be be AUX ajst-10994 133 6 noted note VERB ajst-10994 133 7 that that SCONJ ajst-10994 133 8 on on ADP ajst-10994 133 9 the the DET ajst-10994 133 10 ucoon ucoon PROPN ajst-10994 133 11 dataset dataset NOUN ajst-10994 133 12 , , PUNCT ajst-10994 133 13 the the DET ajst-10994 133 14 tst tst NOUN ajst-10994 133 15 model model NOUN ajst-10994 133 16 exhibits exhibit VERB ajst-10994 133 17 slightly slightly ADV ajst-10994 133 18 different different ADJ ajst-10994 133 19 variations variation NOUN ajst-10994 133 20 in in ADP ajst-10994 133 21 the the DET ajst-10994 133 22 loss loss NOUN ajst-10994 133 23 function function NOUN ajst-10994 133 24 during during ADP ajst-10994 133 25 the the DET ajst-10994 133 26 training training NOUN ajst-10994 133 27 process process NOUN ajst-10994 133 28 compared compare VERB ajst-10994 133 29 to to ADP ajst-10994 133 30 the the DET ajst-10994 133 31 other other ADJ ajst-10994 133 32 two two NUM ajst-10994 133 33 datasets dataset NOUN ajst-10994 133 34 . . PUNCT ajst-10994 134 1 from from ADP ajst-10994 134 2 figure figure NOUN ajst-10994 134 3 6(c 6(c NUM ajst-10994 134 4 ) ) PUNCT ajst-10994 134 5 , , PUNCT ajst-10994 134 6 it it PRON ajst-10994 134 7 can can AUX ajst-10994 134 8 be be AUX ajst-10994 134 9 observed observe VERB ajst-10994 134 10 that that SCONJ ajst-10994 134 11 the the DET ajst-10994 134 12 tst tst NOUN ajst-10994 134 13 model model NOUN ajst-10994 134 14 's 's PART ajst-10994 134 15 loss loss NOUN ajst-10994 134 16 function function NOUN ajst-10994 134 17 on on ADP ajst-10994 134 18 the the DET ajst-10994 134 19 ucoon ucoon PROPN ajst-10994 134 20 dataset dataset PROPN ajst-10994 134 21 shows show VERB ajst-10994 134 22 evident evident ADJ ajst-10994 134 23 periodic periodic ADJ ajst-10994 134 24 decay decay NOUN ajst-10994 134 25 . . PUNCT ajst-10994 135 1 specifically specifically ADV ajst-10994 135 2 , , PUNCT ajst-10994 135 3 in in ADP ajst-10994 135 4 the the DET ajst-10994 135 5 early early ADJ ajst-10994 135 6 stages stage NOUN ajst-10994 135 7 of of ADP ajst-10994 135 8 training training NOUN ajst-10994 135 9 , , PUNCT ajst-10994 135 10 the the DET ajst-10994 135 11 loss loss NOUN ajst-10994 135 12 function function NOUN ajst-10994 135 13 remains remain VERB ajst-10994 135 14 relatively relatively ADV ajst-10994 135 15 unchanged unchanged ADJ ajst-10994 135 16 , , PUNCT ajst-10994 135 17 stabilizing stabilize VERB ajst-10994 135 18 around around ADP ajst-10994 135 19 a a DET ajst-10994 135 20 higher high ADJ ajst-10994 135 21 value value NOUN ajst-10994 135 22 ( ( PUNCT ajst-10994 135 23 0.14 0.14 NUM ajst-10994 135 24 ) ) PUNCT ajst-10994 135 25 until until ADP ajst-10994 135 26 approximately approximately ADV ajst-10994 135 27 8 8 NUM ajst-10994 135 28 steps step NOUN ajst-10994 135 29 of of ADP ajst-10994 135 30 training training NOUN ajst-10994 135 31 when when SCONJ ajst-10994 135 32 a a DET ajst-10994 135 33 more more ADV ajst-10994 135 34 pronounced pronounced ADJ ajst-10994 135 35 decay decay NOUN ajst-10994 135 36 in in ADP ajst-10994 135 37 the the DET ajst-10994 135 38 loss loss NOUN ajst-10994 135 39 function function NOUN ajst-10994 135 40 becomes become VERB ajst-10994 135 41 evident evident ADJ ajst-10994 135 42 . . PUNCT ajst-10994 136 1 these these DET ajst-10994 136 2 results result NOUN ajst-10994 136 3 suggest suggest VERB ajst-10994 136 4 that that SCONJ ajst-10994 136 5 on on ADP ajst-10994 136 6 the the DET ajst-10994 136 7 ucoon ucoon PROPN ajst-10994 136 8 dataset dataset NOUN ajst-10994 136 9 , , PUNCT ajst-10994 136 10 the the DET ajst-10994 136 11 tst tst NOUN ajst-10994 136 12 model model NOUN ajst-10994 136 13 's 's PART ajst-10994 136 14 loss loss NOUN ajst-10994 136 15 function function NOUN ajst-10994 136 16 may may AUX ajst-10994 136 17 encounter encounter VERB ajst-10994 136 18 a a DET ajst-10994 136 19 larger large ADJ ajst-10994 136 20 " " PUNCT ajst-10994 136 21 plateau plateau NOUN ajst-10994 136 22 " " PUNCT ajst-10994 136 23 region region NOUN ajst-10994 136 24 near near ADP ajst-10994 136 25 the the DET ajst-10994 136 26 initial initial ADJ ajst-10994 136 27 parameters parameter NOUN ajst-10994 136 28 , , PUNCT ajst-10994 136 29 which which PRON ajst-10994 136 30 corresponds correspond VERB ajst-10994 136 31 to to ADP ajst-10994 136 32 an an DET ajst-10994 136 33 area area NOUN ajst-10994 136 34 with with ADP ajst-10994 136 35 small small ADJ ajst-10994 136 36 gradients gradient NOUN ajst-10994 136 37 . . PUNCT ajst-10994 137 1 the the DET ajst-10994 137 2 existence existence NOUN ajst-10994 137 3 of of ADP ajst-10994 137 4 this this DET ajst-10994 137 5 region region NOUN ajst-10994 137 6 slows slow VERB ajst-10994 137 7 down down ADP ajst-10994 137 8 the the DET ajst-10994 137 9 convergence convergence NOUN ajst-10994 137 10 speed speed NOUN ajst-10994 137 11 of of ADP ajst-10994 137 12 the the DET ajst-10994 137 13 model model NOUN ajst-10994 137 14 , , PUNCT ajst-10994 137 15 as as SCONJ ajst-10994 137 16 shown show VERB ajst-10994 137 17 in in ADP ajst-10994 137 18 figure figure NOUN ajst-10994 137 19 6(a 6(a NUM ajst-10994 137 20 ) ) PUNCT ajst-10994 137 21 and and CCONJ ajst-10994 137 22 figure figure VERB ajst-10994 137 23 6(b 6(b NUM ajst-10994 137 24 ) ) PUNCT ajst-10994 137 25 . . PUNCT ajst-10994 138 1 unlike unlike ADP ajst-10994 138 2 the the DET ajst-10994 138 3 cwru cwru NOUN ajst-10994 138 4 and and CCONJ ajst-10994 138 5 xjtu xjtu PROPN ajst-10994 138 6 datasets dataset NOUN ajst-10994 138 7 , , PUNCT ajst-10994 138 8 the the DET ajst-10994 138 9 tst tst NOUN ajst-10994 138 10 model model NOUN ajst-10994 138 11 requires require VERB ajst-10994 138 12 about about ADV ajst-10994 138 13 20 20 NUM ajst-10994 138 14 steps step NOUN ajst-10994 138 15 of of ADP ajst-10994 138 16 training training NOUN ajst-10994 138 17 on on ADP ajst-10994 138 18 the the DET ajst-10994 138 19 ucoon ucoon NOUN ajst-10994 138 20 dataset dataset VERB ajst-10994 138 21 to to PART ajst-10994 138 22 gradually gradually ADV ajst-10994 138 23 abolish abolish VERB ajst-10994 138 24 the the DET ajst-10994 138 25 impact impact NOUN ajst-10994 138 26 of of ADP ajst-10994 138 27 haphazard haphazard ADJ ajst-10994 138 28 initialization initialization NOUN ajst-10994 138 29 . . PUNCT ajst-10994 139 1 overall overall ADJ ajst-10994 139 2 , , PUNCT ajst-10994 139 3 at at ADP ajst-10994 139 4 the the DET ajst-10994 139 5 end end NOUN ajst-10994 139 6 of of ADP ajst-10994 139 7 training training NOUN ajst-10994 139 8 , , PUNCT ajst-10994 139 9 the the DET ajst-10994 139 10 tst tst NOUN ajst-10994 139 11 model model NOUN ajst-10994 139 12 achieves achieve VERB ajst-10994 139 13 average average ADJ ajst-10994 139 14 loss loss NOUN ajst-10994 139 15 function function NOUN ajst-10994 139 16 values value NOUN ajst-10994 139 17 of of ADP ajst-10994 139 18 2.95 2.95 NUM ajst-10994 139 19 ร— ร— NOUN ajst-10994 139 20 10โˆ’5 10โˆ’5 NUM ajst-10994 139 21 and and CCONJ ajst-10994 139 22 4.55 4.55 NUM ajst-10994 139 23 ร— ร— NOUN ajst-10994 139 24 10โˆ’4 10โˆ’4 NUM ajst-10994 139 25 on on ADP ajst-10994 139 26 the the DET ajst-10994 139 27 training training NOUN ajst-10994 139 28 set set NOUN ajst-10994 139 29 and and CCONJ ajst-10994 139 30 test test NOUN ajst-10994 139 31 set set NOUN ajst-10994 139 32 of of ADP ajst-10994 139 33 the the DET ajst-10994 139 34 cwru cwru PROPN ajst-10994 139 35 dataset dataset NOUN ajst-10994 139 36 , , PUNCT ajst-10994 139 37 distinctly distinctly ADV ajst-10994 139 38 . . PUNCT ajst-10994 140 1 on on ADP ajst-10994 140 2 the the DET ajst-10994 140 3 xjtu xjtu PROPN ajst-10994 140 4 dataset dataset PROPN ajst-10994 140 5 , , PUNCT ajst-10994 140 6 these these DET ajst-10994 140 7 values value NOUN ajst-10994 140 8 are be AUX ajst-10994 140 9 5.16 5.16 NUM ajst-10994 140 10 ร— ร— NOUN ajst-10994 140 11 10โˆ’6 10โˆ’6 NUM ajst-10994 140 12 and and CCONJ ajst-10994 140 13 5.38 5.38 NUM ajst-10994 140 14 ร— ร— NOUN ajst-10994 140 15 10โˆ’4 10โˆ’4 NUM ajst-10994 140 16 , , PUNCT ajst-10994 140 17 respectively respectively ADV ajst-10994 140 18 . . PUNCT ajst-10994 141 1 on on ADP ajst-10994 141 2 the the DET ajst-10994 141 3 ucoon ucoon PROPN ajst-10994 141 4 dataset dataset NOUN ajst-10994 141 5 , , PUNCT ajst-10994 141 6 the the DET ajst-10994 141 7 values value NOUN ajst-10994 141 8 are be AUX ajst-10994 141 9 3.17 3.17 NUM ajst-10994 141 10 ร— ร— NOUN ajst-10994 141 11 10โˆ’4 10โˆ’4 NUM ajst-10994 141 12 and and CCONJ ajst-10994 141 13 1.51 1.51 NUM ajst-10994 141 14 ร— ร— NOUN ajst-10994 141 15 10โˆ’3 10โˆ’3 NUM ajst-10994 141 16 , , PUNCT ajst-10994 141 17 respectively respectively ADV ajst-10994 141 18 . . PUNCT ajst-10994 142 1 overall overall ADV ajst-10994 142 2 , , PUNCT ajst-10994 142 3 the the DET ajst-10994 142 4 tst tst NOUN ajst-10994 142 5 model model NOUN ajst-10994 142 6 converges converge NOUN ajst-10994 142 7 to to ADP ajst-10994 142 8 relatively relatively ADV ajst-10994 142 9 small small ADJ ajst-10994 142 10 ranges range NOUN ajst-10994 142 11 of of ADP ajst-10994 142 12 loss loss NOUN ajst-10994 142 13 function function NOUN ajst-10994 142 14 values value NOUN ajst-10994 142 15 on on ADP ajst-10994 142 16 all all DET ajst-10994 142 17 three three NUM ajst-10994 142 18 datasets dataset NOUN ajst-10994 142 19 , , PUNCT ajst-10994 142 20 indicating indicate VERB ajst-10994 142 21 that that SCONJ ajst-10994 142 22 the the DET ajst-10994 142 23 tst tst NOUN ajst-10994 142 24 model model NOUN ajst-10994 142 25 exhibits exhibit VERB ajst-10994 142 26 good good ADJ ajst-10994 142 27 convergence convergence NOUN ajst-10994 142 28 characteristics characteristic NOUN ajst-10994 142 29 . . PUNCT ajst-10994 143 1 figure figure NOUN ajst-10994 143 2 5 5 NUM ajst-10994 143 3 . . PUNCT ajst-10994 144 1 changes change NOUN ajst-10994 144 2 in in ADP ajst-10994 144 3 the the DET ajst-10994 144 4 loss loss NOUN ajst-10994 144 5 function function NOUN ajst-10994 144 6 on on ADP ajst-10994 144 7 the the DET ajst-10994 144 8 xjtu xjtu PROPN ajst-10994 144 9 data datum NOUN ajst-10994 144 10 set set VERB ajst-10994 144 11 during during ADP ajst-10994 144 12 the the DET ajst-10994 144 13 training training NOUN ajst-10994 144 14 process process NOUN ajst-10994 144 15 ( ( PUNCT ajst-10994 144 16 a a X ajst-10994 144 17 ) ) PUNCT ajst-10994 144 18 box box NOUN ajst-10994 144 19 diagram diagram NOUN ajst-10994 144 20 of of ADP ajst-10994 144 21 the the DET ajst-10994 144 22 first first ADJ ajst-10994 144 23 10 10 NUM ajst-10994 144 24 generations generation NOUN ajst-10994 144 25 of of ADP ajst-10994 144 26 the the DET ajst-10994 144 27 loss loss NOUN ajst-10994 144 28 function function NOUN ajst-10994 144 29 in in ADP ajst-10994 144 30 the the DET ajst-10994 144 31 training training NOUN ajst-10994 144 32 set set NOUN ajst-10994 144 33 ( ( PUNCT ajst-10994 144 34 b b NOUN ajst-10994 144 35 ) ) PUNCT ajst-10994 144 36 box box NOUN ajst-10994 144 37 diagram diagram NOUN ajst-10994 144 38 of of ADP ajst-10994 144 39 the the DET ajst-10994 144 40 first first ADJ ajst-10994 144 41 10 10 NUM ajst-10994 144 42 generations generation NOUN ajst-10994 144 43 of of ADP ajst-10994 144 44 the the DET ajst-10994 144 45 loss loss NOUN ajst-10994 144 46 function function NOUN ajst-10994 144 47 in in ADP ajst-10994 144 48 the the DET ajst-10994 144 49 test test NOUN ajst-10994 144 50 set set NOUN ajst-10994 144 51 ( ( PUNCT ajst-10994 144 52 c c X ajst-10994 144 53 ) ) PUNCT ajst-10994 144 54 the the DET ajst-10994 144 55 average average ADJ ajst-10994 144 56 loss loss NOUN ajst-10994 144 57 function function NOUN ajst-10994 144 58 value value NOUN ajst-10994 144 59 during during ADP ajst-10994 144 60 the the DET ajst-10994 144 61 entire entire ADJ ajst-10994 144 62 training training NOUN ajst-10994 144 63 process process NOUN ajst-10994 144 64 furthermore furthermore ADV ajst-10994 144 65 , , PUNCT ajst-10994 144 66 figures figure VERB ajst-10994 144 67 7 7 NUM ajst-10994 144 68 to to PART ajst-10994 144 69 9 9 NUM ajst-10994 144 70 illustrate illustrate VERB ajst-10994 144 71 the the DET ajst-10994 144 72 variation variation NOUN ajst-10994 144 73 of of ADP ajst-10994 144 74 fault fault NOUN ajst-10994 144 75 diagnosis diagnosis NOUN ajst-10994 144 76 accuracy accuracy NOUN ajst-10994 144 77 of of ADP ajst-10994 144 78 the the DET ajst-10994 144 79 tst tst NOUN ajst-10994 144 80 ( ( PUNCT ajst-10994 144 81 time time PROPN ajst-10994 144 82 series series PROPN ajst-10994 144 83 transformer transformer PROPN ajst-10994 144 84 ) ) PUNCT ajst-10994 144 85 model model NOUN ajst-10994 144 86 on on ADP ajst-10994 144 87 three three NUM ajst-10994 144 88 datasets dataset NOUN ajst-10994 144 89 during during ADP ajst-10994 144 90 the the DET ajst-10994 144 91 training training NOUN ajst-10994 144 92 process process NOUN ajst-10994 144 93 . . PUNCT ajst-10994 145 1 similar similar ADJ ajst-10994 145 2 to to ADP ajst-10994 145 3 the the DET ajst-10994 145 4 changes change NOUN ajst-10994 145 5 in in ADP ajst-10994 145 6 the the DET ajst-10994 145 7 loss loss NOUN ajst-10994 145 8 function function NOUN ajst-10994 145 9 , , PUNCT ajst-10994 145 10 the the DET ajst-10994 145 11 results result NOUN ajst-10994 145 12 are be AUX ajst-10994 145 13 presented present VERB ajst-10994 145 14 in in ADP ajst-10994 145 15 two two NUM ajst-10994 145 16 forms form NOUN ajst-10994 145 17 : : PUNCT ajst-10994 145 18 boxplots boxplot NOUN ajst-10994 145 19 representing represent VERB ajst-10994 145 20 early early ADJ ajst-10994 145 21 training training NOUN ajst-10994 145 22 stages stage NOUN ajst-10994 145 23 and and CCONJ ajst-10994 145 24 the the DET ajst-10994 145 25 average average ADJ ajst-10994 145 26 values value NOUN ajst-10994 145 27 throughout throughout ADP ajst-10994 145 28 the the DET ajst-10994 145 29 entire entire ADJ ajst-10994 145 30 training training NOUN ajst-10994 145 31 process process NOUN ajst-10994 145 32 . . PUNCT ajst-10994 146 1 on on ADP ajst-10994 146 2 the the DET ajst-10994 146 3 cwru cwru NOUN ajst-10994 146 4 dataset dataset NOUN ajst-10994 146 5 , , PUNCT ajst-10994 146 6 as as SCONJ ajst-10994 146 7 shown show VERB ajst-10994 146 8 in in ADP ajst-10994 146 9 figures figure NOUN ajst-10994 146 10 7(a 7(a NUM ajst-10994 146 11 ) ) PUNCT ajst-10994 146 12 and and CCONJ ajst-10994 146 13 7(b 7(b X ajst-10994 146 14 ) ) PUNCT ajst-10994 146 15 , , PUNCT ajst-10994 146 16 the the DET ajst-10994 146 17 accuracy accuracy NOUN ajst-10994 146 18 of of ADP ajst-10994 146 19 the the DET ajst-10994 146 20 tst tst NOUN ajst-10994 146 21 model model NOUN ajst-10994 146 22 on on ADP ajst-10994 146 23 the the DET ajst-10994 146 24 training training NOUN ajst-10994 146 25 and and CCONJ ajst-10994 146 26 test test NOUN ajst-10994 146 27 sets set NOUN ajst-10994 146 28 fluctuates fluctuate NOUN ajst-10994 146 29 significantly significantly ADV ajst-10994 146 30 during during ADP ajst-10994 146 31 the the DET ajst-10994 146 32 early early ADJ ajst-10994 146 33 stages stage NOUN ajst-10994 146 34 of of ADP ajst-10994 146 35 training training NOUN ajst-10994 146 36 , , PUNCT ajst-10994 146 37 likely likely ADJ ajst-10994 146 38 due due ADP ajst-10994 146 39 to to ADP ajst-10994 146 40 the the DET ajst-10994 146 41 impact impact NOUN ajst-10994 146 42 of of ADP ajst-10994 146 43 random random ADJ ajst-10994 146 44 initialization initialization NOUN ajst-10994 146 45 . . PUNCT ajst-10994 147 1 however however ADV ajst-10994 147 2 , , PUNCT ajst-10994 147 3 after after ADP ajst-10994 147 4 approximately approximately ADV ajst-10994 147 5 10 10 NUM ajst-10994 147 6 training training NOUN ajst-10994 147 7 steps step NOUN ajst-10994 147 8 , , PUNCT ajst-10994 147 9 the the DET ajst-10994 147 10 shape shape NOUN ajst-10994 147 11 of of ADP ajst-10994 147 12 the the DET ajst-10994 147 13 boxplots boxplot NOUN ajst-10994 147 14 gradually gradually ADV ajst-10994 147 15 becomes become VERB ajst-10994 147 16 flat flat ADJ ajst-10994 147 17 , , PUNCT ajst-10994 147 18 indicating indicate VERB ajst-10994 147 19 that that SCONJ ajst-10994 147 20 the the DET ajst-10994 147 21 influence influence NOUN ajst-10994 147 22 of of ADP ajst-10994 147 23 random random ADJ ajst-10994 147 24 initialization initialization NOUN ajst-10994 147 25 has have AUX ajst-10994 147 26 been be AUX ajst-10994 147 27 mostly mostly ADV ajst-10994 147 28 eliminated eliminate VERB ajst-10994 147 29 , , PUNCT ajst-10994 147 30 and and CCONJ ajst-10994 147 31 the the DET ajst-10994 147 32 recognition recognition NOUN ajst-10994 147 33 accuracy accuracy NOUN ajst-10994 147 34 stabilizes stabilize VERB ajst-10994 147 35 as as SCONJ ajst-10994 147 36 the the DET ajst-10994 147 37 training training NOUN ajst-10994 147 38 progresses progress VERB ajst-10994 147 39 . . PUNCT ajst-10994 148 1 moreover moreover ADV ajst-10994 148 2 , , PUNCT ajst-10994 148 3 as as SCONJ ajst-10994 148 4 depicted depict VERB ajst-10994 148 5 in in ADP ajst-10994 148 6 figure figure NOUN ajst-10994 148 7 7(c 7(c NUM ajst-10994 148 8 ) ) PUNCT ajst-10994 148 9 , , PUNCT ajst-10994 148 10 it it PRON ajst-10994 148 11 can can AUX ajst-10994 148 12 be be AUX ajst-10994 148 13 observed observe VERB ajst-10994 148 14 that that SCONJ ajst-10994 148 15 the the DET ajst-10994 148 16 average average ADJ ajst-10994 148 17 accuracy accuracy NOUN ajst-10994 148 18 of of ADP ajst-10994 148 19 the the DET ajst-10994 148 20 tst tst NOUN ajst-10994 148 21 model model NOUN ajst-10994 148 22 on on ADP ajst-10994 148 23 both both CCONJ ajst-10994 148 24 the the DET ajst-10994 148 25 training training NOUN ajst-10994 148 26 and and CCONJ ajst-10994 148 27 test test NOUN ajst-10994 148 28 sets set NOUN ajst-10994 148 29 increases increase NOUN ajst-10994 148 30 steadily steadily ADV ajst-10994 148 31 throughout throughout ADP ajst-10994 148 32 the the DET ajst-10994 148 33 entire entire ADJ ajst-10994 148 34 training training NOUN ajst-10994 148 35 process process NOUN ajst-10994 148 36 . . PUNCT ajst-10994 149 1 at at ADP ajst-10994 149 2 the the DET ajst-10994 149 3 end end NOUN ajst-10994 149 4 of of ADP ajst-10994 149 5 training training NOUN ajst-10994 149 6 , , PUNCT ajst-10994 149 7 the the DET ajst-10994 149 8 tst tst NOUN ajst-10994 149 9 model model NOUN ajst-10994 149 10 achieves achieve VERB ajst-10994 149 11 high high ADJ ajst-10994 149 12 recognition recognition NOUN ajst-10994 149 13 accuracy accuracy NOUN ajst-10994 149 14 on on ADP ajst-10994 149 15 both both CCONJ ajst-10994 149 16 the the DET ajst-10994 149 17 training training NOUN ajst-10994 149 18 and and CCONJ ajst-10994 149 19 test test NOUN ajst-10994 149 20 sets set NOUN ajst-10994 149 21 , , PUNCT ajst-10994 149 22 demonstrating demonstrate VERB ajst-10994 149 23 that that SCONJ ajst-10994 149 24 the the DET ajst-10994 149 25 approach approach NOUN ajst-10994 149 26 proposed propose VERB ajst-10994 149 27 in in ADP ajst-10994 149 28 this this DET ajst-10994 149 29 chapter chapter NOUN ajst-10994 149 30 can can AUX ajst-10994 149 31 achieve achieve VERB ajst-10994 149 32 accurate accurate ADJ ajst-10994 149 33 fault fault NOUN ajst-10994 149 34 diagnosis diagnosis NOUN ajst-10994 149 35 . . PUNCT ajst-10994 150 1 84 84 NUM ajst-10994 150 2 figure figure NOUN ajst-10994 150 3 6 6 NUM ajst-10994 150 4 . . PUNCT ajst-10994 150 5 loss loss NOUN ajst-10994 150 6 function function NOUN ajst-10994 150 7 changes change NOUN ajst-10994 150 8 on on ADP ajst-10994 150 9 ucoon ucoon PROPN ajst-10994 150 10 data datum NOUN ajst-10994 150 11 set set VERB ajst-10994 150 12 during during ADP ajst-10994 150 13 the the DET ajst-10994 150 14 training training NOUN ajst-10994 150 15 process process NOUN ajst-10994 150 16 ( ( PUNCT ajst-10994 150 17 a a X ajst-10994 150 18 ) ) PUNCT ajst-10994 150 19 box box NOUN ajst-10994 150 20 diagram diagram NOUN ajst-10994 150 21 of of ADP ajst-10994 150 22 the the DET ajst-10994 150 23 first first ADJ ajst-10994 150 24 20 20 NUM ajst-10994 150 25 generations generation NOUN ajst-10994 150 26 of of ADP ajst-10994 150 27 loss loss NOUN ajst-10994 150 28 functions function NOUN ajst-10994 150 29 in in ADP ajst-10994 150 30 the the DET ajst-10994 150 31 training training NOUN ajst-10994 150 32 set set NOUN ajst-10994 150 33 ( ( PUNCT ajst-10994 150 34 b b NOUN ajst-10994 150 35 ) ) PUNCT ajst-10994 150 36 box box NOUN ajst-10994 150 37 diagram diagram NOUN ajst-10994 150 38 of of ADP ajst-10994 150 39 the the DET ajst-10994 150 40 first first ADJ ajst-10994 150 41 20 20 NUM ajst-10994 150 42 generations generation NOUN ajst-10994 150 43 of of ADP ajst-10994 150 44 loss loss NOUN ajst-10994 150 45 functions function NOUN ajst-10994 150 46 in in ADP ajst-10994 150 47 the the DET ajst-10994 150 48 test test NOUN ajst-10994 150 49 set set NOUN ajst-10994 150 50 ( ( PUNCT ajst-10994 150 51 c c NOUN ajst-10994 150 52 ) ) PUNCT ajst-10994 150 53 average average ADJ ajst-10994 150 54 loss loss NOUN ajst-10994 150 55 function function NOUN ajst-10994 150 56 values value NOUN ajst-10994 150 57 during during ADP ajst-10994 150 58 the the DET ajst-10994 150 59 entire entire ADJ ajst-10994 150 60 training training NOUN ajst-10994 150 61 process process NOUN ajst-10994 150 62 figure figure NOUN ajst-10994 150 63 7 7 NUM ajst-10994 150 64 . . PUNCT ajst-10994 150 65 accuracy accuracy NOUN ajst-10994 150 66 changes change NOUN ajst-10994 150 67 on on ADP ajst-10994 150 68 the the DET ajst-10994 150 69 cwru cwru PROPN ajst-10994 150 70 data datum NOUN ajst-10994 150 71 set set VERB ajst-10994 150 72 during during ADP ajst-10994 150 73 the the DET ajst-10994 150 74 training training NOUN ajst-10994 150 75 process process NOUN ajst-10994 150 76 ( ( PUNCT ajst-10994 150 77 a a X ajst-10994 150 78 ) ) PUNCT ajst-10994 150 79 box box NOUN ajst-10994 150 80 diagram diagram NOUN ajst-10994 150 81 of of ADP ajst-10994 150 82 the the DET ajst-10994 150 83 first first ADJ ajst-10994 150 84 10 10 NUM ajst-10994 150 85 generations generation NOUN ajst-10994 150 86 of of ADP ajst-10994 150 87 accuracy accuracy NOUN ajst-10994 150 88 in in ADP ajst-10994 150 89 the the DET ajst-10994 150 90 training training NOUN ajst-10994 150 91 set set NOUN ajst-10994 150 92 ( ( PUNCT ajst-10994 150 93 b b NOUN ajst-10994 150 94 ) ) PUNCT ajst-10994 150 95 box box NOUN ajst-10994 150 96 diagram diagram NOUN ajst-10994 150 97 of of ADP ajst-10994 150 98 the the DET ajst-10994 150 99 first first ADJ ajst-10994 150 100 10 10 NUM ajst-10994 150 101 generations generation NOUN ajst-10994 150 102 of of ADP ajst-10994 150 103 accuracy accuracy NOUN ajst-10994 150 104 in in ADP ajst-10994 150 105 the the DET ajst-10994 150 106 test test NOUN ajst-10994 150 107 set set NOUN ajst-10994 150 108 ( ( PUNCT ajst-10994 150 109 c c NOUN ajst-10994 150 110 ) ) PUNCT ajst-10994 150 111 mean mean ADJ ajst-10994 150 112 achievement achievement NOUN ajst-10994 150 113 throughout throughout ADP ajst-10994 150 114 training training NOUN ajst-10994 150 115 figure figure NOUN ajst-10994 150 116 8 8 NUM ajst-10994 150 117 . . PUNCT ajst-10994 151 1 accuracy accuracy NOUN ajst-10994 151 2 changes change NOUN ajst-10994 151 3 on on ADP ajst-10994 151 4 the the DET ajst-10994 151 5 xjtu xjtu PROPN ajst-10994 151 6 data datum NOUN ajst-10994 151 7 set set VERB ajst-10994 151 8 during during ADP ajst-10994 151 9 training training NOUN ajst-10994 151 10 ( ( PUNCT ajst-10994 151 11 a a PRON ajst-10994 151 12 ) ) PUNCT ajst-10994 151 13 box box NOUN ajst-10994 151 14 diagram diagram NOUN ajst-10994 151 15 of of ADP ajst-10994 151 16 the the DET ajst-10994 151 17 first first ADJ ajst-10994 151 18 10 10 NUM ajst-10994 151 19 generations generation NOUN ajst-10994 151 20 of of ADP ajst-10994 151 21 accuracy accuracy NOUN ajst-10994 151 22 in in ADP ajst-10994 151 23 the the DET ajst-10994 151 24 training training NOUN ajst-10994 151 25 set set NOUN ajst-10994 151 26 ( ( PUNCT ajst-10994 151 27 b b NOUN ajst-10994 151 28 ) ) PUNCT ajst-10994 151 29 box box NOUN ajst-10994 151 30 diagram diagram NOUN ajst-10994 151 31 of of ADP ajst-10994 151 32 the the DET ajst-10994 151 33 first first ADJ ajst-10994 151 34 10 10 NUM ajst-10994 151 35 generations generation NOUN ajst-10994 151 36 of of ADP ajst-10994 151 37 accuracy accuracy NOUN ajst-10994 151 38 in in ADP ajst-10994 151 39 the the DET ajst-10994 151 40 test test NOUN ajst-10994 151 41 set set NOUN ajst-10994 151 42 ( ( PUNCT ajst-10994 151 43 c c NOUN ajst-10994 151 44 ) ) PUNCT ajst-10994 151 45 mean mean NOUN ajst-10994 151 46 success success NOUN ajst-10994 151 47 rate rate NOUN ajst-10994 151 48 during during ADP ajst-10994 151 49 training training NOUN ajst-10994 151 50 iterations iteration NOUN ajst-10994 151 51 figure figure NOUN ajst-10994 151 52 9 9 NUM ajst-10994 151 53 . . PUNCT ajst-10994 152 1 accuracy accuracy NOUN ajst-10994 152 2 changes change NOUN ajst-10994 152 3 on on ADP ajst-10994 152 4 the the DET ajst-10994 152 5 ucoon ucoon PROPN ajst-10994 152 6 data datum NOUN ajst-10994 152 7 set set VERB ajst-10994 152 8 during during ADP ajst-10994 152 9 the the DET ajst-10994 152 10 training training NOUN ajst-10994 152 11 process process NOUN ajst-10994 152 12 ( ( PUNCT ajst-10994 152 13 a a X ajst-10994 152 14 ) ) PUNCT ajst-10994 152 15 box box NOUN ajst-10994 152 16 diagram diagram NOUN ajst-10994 152 17 of of ADP ajst-10994 152 18 the the DET ajst-10994 152 19 first first ADJ ajst-10994 152 20 20 20 NUM ajst-10994 152 21 generations generation NOUN ajst-10994 152 22 of of ADP ajst-10994 152 23 accuracy accuracy NOUN ajst-10994 152 24 in in ADP ajst-10994 152 25 the the DET ajst-10994 152 26 training training NOUN ajst-10994 152 27 set set NOUN ajst-10994 152 28 ( ( PUNCT ajst-10994 152 29 b b NOUN ajst-10994 152 30 ) ) PUNCT ajst-10994 152 31 box box NOUN ajst-10994 152 32 diagram diagram NOUN ajst-10994 152 33 of of ADP ajst-10994 152 34 the the DET ajst-10994 152 35 first first ADJ ajst-10994 152 36 20 20 NUM ajst-10994 152 37 generations generation NOUN ajst-10994 152 38 of of ADP ajst-10994 152 39 accuracy accuracy NOUN ajst-10994 152 40 in in ADP ajst-10994 152 41 the the DET ajst-10994 152 42 test test NOUN ajst-10994 152 43 set set NOUN ajst-10994 152 44 ( ( PUNCT ajst-10994 152 45 c c NOUN ajst-10994 152 46 ) ) PUNCT ajst-10994 152 47 average average ADJ ajst-10994 152 48 accuracy accuracy NOUN ajst-10994 152 49 during during ADP ajst-10994 152 50 the the DET ajst-10994 152 51 entire entire ADJ ajst-10994 152 52 training training NOUN ajst-10994 152 53 process process NOUN ajst-10994 152 54 similar similar ADJ ajst-10994 152 55 trends trend NOUN ajst-10994 152 56 are be AUX ajst-10994 152 57 also also ADV ajst-10994 152 58 observed observe VERB ajst-10994 152 59 in in ADP ajst-10994 152 60 the the DET ajst-10994 152 61 results result NOUN ajst-10994 152 62 presented present VERB ajst-10994 152 63 in in ADP ajst-10994 152 64 figures figure NOUN ajst-10994 152 65 8 8 NUM ajst-10994 152 66 and and CCONJ ajst-10994 152 67 9 9 NUM ajst-10994 152 68 . . PUNCT ajst-10994 153 1 additionally additionally ADV ajst-10994 153 2 , , PUNCT ajst-10994 153 3 it it PRON ajst-10994 153 4 is be AUX ajst-10994 153 5 essential essential ADJ ajst-10994 153 6 to to PART ajst-10994 153 7 note note VERB ajst-10994 153 8 that that SCONJ ajst-10994 153 9 on on ADP ajst-10994 153 10 the the DET ajst-10994 153 11 ucoon ucoon PROPN ajst-10994 153 12 dataset dataset NOUN ajst-10994 153 13 , , PUNCT ajst-10994 153 14 homogeneous homogeneous ADJ ajst-10994 153 15 to to ADP ajst-10994 153 16 the the DET ajst-10994 153 17 changes change NOUN ajst-10994 153 18 within within ADP ajst-10994 153 19 the the DET ajst-10994 153 20 loss loss NOUN ajst-10994 153 21 function function NOUN ajst-10994 153 22 , , PUNCT ajst-10994 153 23 the the DET ajst-10994 153 24 tst tst NOUN ajst-10994 153 25 model model NOUN ajst-10994 153 26 's 's PART ajst-10994 153 27 accuracy accuracy NOUN ajst-10994 153 28 exhibits exhibit VERB ajst-10994 153 29 a a DET ajst-10994 153 30 distinct distinct ADJ ajst-10994 153 31 flat flat ADJ ajst-10994 153 32 phase phase NOUN ajst-10994 153 33 during during ADP ajst-10994 153 34 the the DET ajst-10994 153 35 early early ADJ ajst-10994 153 36 stages stage NOUN ajst-10994 153 37 of of ADP ajst-10994 153 38 training training NOUN ajst-10994 153 39 . . PUNCT ajst-10994 154 1 the the DET ajst-10994 154 2 recognition recognition NOUN ajst-10994 154 3 accuracy accuracy NOUN ajst-10994 154 4 on on ADP ajst-10994 154 5 both both CCONJ ajst-10994 154 6 the the DET ajst-10994 154 7 training training NOUN ajst-10994 154 8 and and CCONJ ajst-10994 154 9 test test NOUN ajst-10994 154 10 sets set NOUN ajst-10994 154 11 remains remain VERB ajst-10994 154 12 around around ADP ajst-10994 154 13 11 11 NUM ajst-10994 154 14 % % NOUN ajst-10994 154 15 , , PUNCT ajst-10994 154 16 which which PRON ajst-10994 154 17 is be AUX ajst-10994 154 18 approximately approximately ADV ajst-10994 154 19 equivalent equivalent ADJ ajst-10994 154 20 to to ADP ajst-10994 154 21 random random ADJ ajst-10994 154 22 guessing guessing NOUN ajst-10994 154 23 . . PUNCT ajst-10994 155 1 this this PRON ajst-10994 155 2 indicates indicate VERB ajst-10994 155 3 that that SCONJ ajst-10994 155 4 the the DET ajst-10994 155 5 model model NOUN ajst-10994 155 6 did do AUX ajst-10994 155 7 not not PART ajst-10994 155 8 undergo undergo VERB ajst-10994 155 9 effective effective ADJ ajst-10994 155 10 parameter parameter NOUN ajst-10994 155 11 updates update NOUN ajst-10994 155 12 during during ADP ajst-10994 155 13 the the DET ajst-10994 155 14 initial initial ADJ ajst-10994 155 15 training training NOUN ajst-10994 155 16 stage stage NOUN ajst-10994 155 17 , , PUNCT ajst-10994 155 18 further far ADV ajst-10994 155 19 confirming confirm VERB ajst-10994 155 20 the the DET ajst-10994 155 21 existence existence NOUN ajst-10994 155 22 of of ADP ajst-10994 155 23 a a DET ajst-10994 155 24 " " PUNCT ajst-10994 155 25 plateau plateau NOUN ajst-10994 155 26 region region NOUN ajst-10994 155 27 " " PUNCT ajst-10994 155 28 around around ADP ajst-10994 155 29 the the DET ajst-10994 155 30 initial initial ADJ ajst-10994 155 31 values value NOUN ajst-10994 155 32 . . PUNCT ajst-10994 156 1 overall overall ADV ajst-10994 156 2 , , PUNCT ajst-10994 156 3 when when SCONJ ajst-10994 156 4 training training NOUN ajst-10994 156 5 concludes conclude VERB ajst-10994 156 6 , , PUNCT ajst-10994 156 7 the the DET ajst-10994 156 8 tst tst NOUN ajst-10994 156 9 model model NOUN ajst-10994 156 10 achieves achieve VERB ajst-10994 156 11 average average ADJ ajst-10994 156 12 accuracy accuracy NOUN ajst-10994 156 13 of of ADP ajst-10994 156 14 100 100 NUM ajst-10994 156 15 % % NOUN ajst-10994 156 16 on on ADP ajst-10994 156 17 the the DET ajst-10994 156 18 training training NOUN ajst-10994 156 19 set set NOUN ajst-10994 156 20 and and CCONJ ajst-10994 156 21 98.63 98.63 NUM ajst-10994 156 22 % % NOUN ajst-10994 156 23 on on ADP ajst-10994 156 24 the the DET ajst-10994 156 25 test test NOUN ajst-10994 156 26 set set VERB ajst-10994 156 27 for for ADP ajst-10994 156 28 the the DET ajst-10994 156 29 cwru cwru PROPN ajst-10994 156 30 dataset dataset NOUN ajst-10994 156 31 , , PUNCT ajst-10994 156 32 100 100 NUM ajst-10994 156 33 % % NOUN ajst-10994 156 34 on on ADP ajst-10994 156 35 the the DET ajst-10994 156 36 training training NOUN ajst-10994 156 37 set set NOUN ajst-10994 156 38 and and CCONJ ajst-10994 156 39 99.78 99.78 NUM ajst-10994 156 40 % % NOUN ajst-10994 156 41 on on ADP ajst-10994 156 42 the the DET ajst-10994 156 43 test test NOUN ajst-10994 156 44 set set VERB ajst-10994 156 45 for for ADP ajst-10994 156 46 the the DET ajst-10994 156 47 xjtu xjtu PROPN ajst-10994 156 48 dataset dataset PROPN ajst-10994 156 49 , , PUNCT ajst-10994 156 50 and and CCONJ ajst-10994 156 51 100 100 NUM ajst-10994 156 52 % % NOUN ajst-10994 156 53 on on ADP ajst-10994 156 54 the the DET ajst-10994 156 55 training training NOUN ajst-10994 156 56 set set NOUN ajst-10994 156 57 and and CCONJ ajst-10994 156 58 99.51 99.51 NUM ajst-10994 156 59 % % NOUN ajst-10994 156 60 on on ADP ajst-10994 156 61 the the DET ajst-10994 156 62 test test NOUN ajst-10994 156 63 set set VERB ajst-10994 156 64 for for ADP ajst-10994 156 65 the the DET ajst-10994 156 66 ucoon ucoon PROPN ajst-10994 156 67 dataset dataset NOUN ajst-10994 156 68 . . PUNCT ajst-10994 157 1 consequently consequently ADV ajst-10994 157 2 , , PUNCT ajst-10994 157 3 the the DET ajst-10994 157 4 tst tst NOUN ajst-10994 157 5 model model NOUN ajst-10994 157 6 attains attain VERB ajst-10994 157 7 an an DET ajst-10994 157 8 average average ADJ ajst-10994 157 9 fault fault NOUN ajst-10994 157 10 diagnosis diagnosis NOUN ajst-10994 157 11 accuracy accuracy NOUN ajst-10994 157 12 of of ADP ajst-10994 157 13 over over ADP ajst-10994 157 14 95 95 NUM ajst-10994 157 15 % % NOUN ajst-10994 157 16 across across ADP ajst-10994 157 17 all all DET ajst-10994 157 18 three three NUM ajst-10994 157 19 datasets dataset NOUN ajst-10994 157 20 , , PUNCT ajst-10994 157 21 demonstrating demonstrate VERB ajst-10994 157 22 its its PRON ajst-10994 157 23 excellent excellent ADJ ajst-10994 157 24 fault fault NOUN ajst-10994 157 25 diagnosis diagnosis NOUN ajst-10994 157 26 performance performance NOUN ajst-10994 157 27 . . PUNCT ajst-10994 158 1 4.2 4.2 NUM ajst-10994 158 2 . . PUNCT ajst-10994 158 3 comparison comparison NOUN ajst-10994 158 4 with with ADP ajst-10994 158 5 other other ADJ ajst-10994 158 6 fault fault NOUN ajst-10994 158 7 diagnosis diagnosis NOUN ajst-10994 158 8 methods method NOUN ajst-10994 158 9 it it PRON ajst-10994 158 10 can can AUX ajst-10994 158 11 be be AUX ajst-10994 158 12 seen see VERB ajst-10994 158 13 from from ADP ajst-10994 158 14 the the DET ajst-10994 158 15 results result NOUN ajst-10994 158 16 in in ADP ajst-10994 158 17 the the DET ajst-10994 158 18 table table NOUN ajst-10994 158 19 that that SCONJ ajst-10994 158 20 the the DET ajst-10994 158 21 tst tst NOUN ajst-10994 158 22 model model NOUN ajst-10994 158 23 proposed propose VERB ajst-10994 158 24 in in ADP ajst-10994 158 25 this this DET ajst-10994 158 26 paper paper NOUN ajst-10994 158 27 can can AUX ajst-10994 158 28 achieve achieve VERB ajst-10994 158 29 a a DET ajst-10994 158 30 higher high ADJ ajst-10994 158 31 accuracy accuracy NOUN ajst-10994 158 32 than than ADP ajst-10994 158 33 many many ADJ ajst-10994 158 34 existing exist VERB ajst-10994 158 35 fault fault NOUN ajst-10994 158 36 diagnosis diagnosis NOUN ajst-10994 158 37 methods method NOUN ajst-10994 158 38 without without ADP ajst-10994 158 39 any any DET ajst-10994 158 40 85 85 NUM ajst-10994 158 41 additional additional ADJ ajst-10994 158 42 preprocessing preprocessing NOUN ajst-10994 158 43 , , PUNCT ajst-10994 158 44 which which PRON ajst-10994 158 45 further far ADV ajst-10994 158 46 proves prove VERB ajst-10994 158 47 the the DET ajst-10994 158 48 effectiveness effectiveness NOUN ajst-10994 158 49 and and CCONJ ajst-10994 158 50 superiority superiority NOUN ajst-10994 158 51 of of ADP ajst-10994 158 52 the the DET ajst-10994 158 53 proposed propose VERB ajst-10994 158 54 method method NOUN ajst-10994 158 55 . . PUNCT ajst-10994 159 1 table table NOUN ajst-10994 159 2 1 1 NUM ajst-10994 159 3 . . PUNCT ajst-10994 160 1 results result NOUN ajst-10994 160 2 of of ADP ajst-10994 160 3 comparison comparison NOUN ajst-10994 160 4 between between ADP ajst-10994 160 5 tst tst NOUN ajst-10994 160 6 model model NOUN ajst-10994 160 7 and and CCONJ ajst-10994 160 8 other other ADJ ajst-10994 160 9 existing exist VERB ajst-10994 160 10 fault fault NOUN ajst-10994 160 11 diagnosis diagnosis NOUN ajst-10994 160 12 methods method NOUN ajst-10994 160 13 model model NOUN ajst-10994 160 14 name name NOUN ajst-10994 160 15 signal signal NOUN ajst-10994 160 16 processing processing NOUN ajst-10994 160 17 methods method NOUN ajst-10994 160 18 and and CCONJ ajst-10994 160 19 others other NOUN ajst-10994 160 20 accuracy accuracy NOUN ajst-10994 160 21 rate rate NOUN ajst-10994 160 22 tst tst NOUN ajst-10994 160 23 ( ( PUNCT ajst-10994 160 24 proposed propose VERB ajst-10994 160 25 ) ) PUNCT ajst-10994 160 26 raw raw ADJ ajst-10994 160 27 vibration vibration NOUN ajst-10994 160 28 signal signal NOUN ajst-10994 160 29 data datum NOUN ajst-10994 160 30 99.72 99.72 NUM ajst-10994 160 31 % % NOUN ajst-10994 160 32 nkh nkh ADV ajst-10994 160 33 - - PUNCT ajst-10994 160 34 kelm kelm NOUN ajst-10994 160 35 multiscale multiscale ADJ ajst-10994 160 36 dispersion dispersion NOUN ajst-10994 160 37 entropy entropy NOUN ajst-10994 160 38 95.56 95.56 NUM ajst-10994 160 39 % % NOUN ajst-10994 160 40 cwt cwt PROPN ajst-10994 160 41 - - PUNCT ajst-10994 160 42 cnn cnn PROPN ajst-10994 160 43 time time NOUN ajst-10994 160 44 - - PUNCT ajst-10994 160 45 frequency frequency NOUN ajst-10994 160 46 graph graph NOUN ajst-10994 160 47 obtained obtain VERB ajst-10994 160 48 by by ADP ajst-10994 160 49 wavelet wavelet NOUN ajst-10994 160 50 transform transform VERB ajst-10994 160 51 99.40 99.40 NUM ajst-10994 160 52 % % NOUN ajst-10994 160 53 dcn dcn PROPN ajst-10994 160 54 raw raw ADJ ajst-10994 160 55 vibration vibration NOUN ajst-10994 160 56 signal signal NOUN ajst-10994 160 57 data datum NOUN ajst-10994 160 58 99.31 99.31 NUM ajst-10994 160 59 % % NOUN ajst-10994 160 60 lstm lstm ADJ ajst-10994 160 61 short short ADJ ajst-10994 160 62 - - PUNCT ajst-10994 160 63 time time NOUN ajst-10994 160 64 fourier fourier NOUN ajst-10994 160 65 transform transform VERB ajst-10994 160 66 98.65 98.65 NUM ajst-10994 160 67 % % NOUN ajst-10994 160 68 5 5 NUM ajst-10994 160 69 . . PUNCT ajst-10994 160 70 conclusion conclusion NOUN ajst-10994 160 71 this this DET ajst-10994 160 72 paper paper NOUN ajst-10994 160 73 proposes propose VERB ajst-10994 160 74 an an DET ajst-10994 160 75 innovative innovative ADJ ajst-10994 160 76 fault fault NOUN ajst-10994 160 77 diagnosis diagnosis NOUN ajst-10994 160 78 model model NOUN ajst-10994 160 79 called call VERB ajst-10994 160 80 time time NOUN ajst-10994 160 81 series series PROPN ajst-10994 160 82 transformer transformer NOUN ajst-10994 160 83 ( ( PUNCT ajst-10994 160 84 tst tst NOUN ajst-10994 160 85 ) ) PUNCT ajst-10994 160 86 and and CCONJ ajst-10994 160 87 demonstrates demonstrate VERB ajst-10994 160 88 its its PRON ajst-10994 160 89 effectiveness effectiveness NOUN ajst-10994 160 90 in in ADP ajst-10994 160 91 the the DET ajst-10994 160 92 field field NOUN ajst-10994 160 93 of of ADP ajst-10994 160 94 fault fault NOUN ajst-10994 160 95 diagnosis diagnosis NOUN ajst-10994 160 96 through through ADP ajst-10994 160 97 validation validation NOUN ajst-10994 160 98 on on ADP ajst-10994 160 99 three three NUM ajst-10994 160 100 experimental experimental ADJ ajst-10994 160 101 datasets dataset NOUN ajst-10994 160 102 : : PUNCT ajst-10994 160 103 cwru cwru PROPN ajst-10994 160 104 , , PUNCT ajst-10994 160 105 xjtu xjtu PROPN ajst-10994 160 106 , , PUNCT ajst-10994 160 107 and and CCONJ ajst-10994 160 108 ucoon ucoon NOUN ajst-10994 160 109 . . PUNCT ajst-10994 161 1 compared compare VERB ajst-10994 161 2 to to ADP ajst-10994 161 3 existing exist VERB ajst-10994 161 4 fault fault NOUN ajst-10994 161 5 diagnosis diagnosis NOUN ajst-10994 161 6 methods method NOUN ajst-10994 161 7 , , PUNCT ajst-10994 161 8 the the DET ajst-10994 161 9 tst tst NOUN ajst-10994 161 10 model model NOUN ajst-10994 161 11 shows show VERB ajst-10994 161 12 significant significant ADJ ajst-10994 161 13 advantages advantage NOUN ajst-10994 161 14 . . PUNCT ajst-10994 162 1 overall overall ADV ajst-10994 162 2 , , PUNCT ajst-10994 162 3 this this DET ajst-10994 162 4 paper paper NOUN ajst-10994 162 5 makes make VERB ajst-10994 162 6 a a DET ajst-10994 162 7 valuable valuable ADJ ajst-10994 162 8 contribution contribution NOUN ajst-10994 162 9 to to ADP ajst-10994 162 10 the the DET ajst-10994 162 11 field field NOUN ajst-10994 162 12 of of ADP ajst-10994 162 13 fault fault NOUN ajst-10994 162 14 diagnosis diagnosis NOUN ajst-10994 162 15 by by ADP ajst-10994 162 16 introducing introduce VERB ajst-10994 162 17 the the DET ajst-10994 162 18 time time NOUN ajst-10994 162 19 series series PROPN ajst-10994 162 20 transformer transformer NOUN ajst-10994 162 21 model model NOUN ajst-10994 162 22 and and CCONJ ajst-10994 162 23 effectively effectively ADV ajst-10994 162 24 addressing address VERB ajst-10994 162 25 various various ADJ ajst-10994 162 26 fault fault NOUN ajst-10994 162 27 diagnosis diagnosis NOUN ajst-10994 162 28 issues issue NOUN ajst-10994 162 29 . . PUNCT ajst-10994 163 1 future future ADJ ajst-10994 163 2 work work NOUN ajst-10994 163 3 can can AUX ajst-10994 163 4 build build VERB ajst-10994 163 5 upon upon SCONJ ajst-10994 163 6 this this DET ajst-10994 163 7 foundation foundation NOUN ajst-10994 163 8 to to PART ajst-10994 163 9 further far ADV ajst-10994 163 10 refine refine VERB ajst-10994 163 11 and and CCONJ ajst-10994 163 12 expand expand VERB ajst-10994 163 13 the the DET ajst-10994 163 14 model model NOUN ajst-10994 163 15 , , PUNCT ajst-10994 163 16 enabling enable VERB ajst-10994 163 17 it it PRON ajst-10994 163 18 to to PART ajst-10994 163 19 better well ADV ajst-10994 163 20 handle handle VERB ajst-10994 163 21 practical practical ADJ ajst-10994 163 22 engineering engineering NOUN ajst-10994 163 23 fault fault NOUN ajst-10994 163 24 diagnosis diagnosis NOUN ajst-10994 163 25 challenges challenge NOUN ajst-10994 163 26 . . PUNCT ajst-10994 164 1 additionally additionally ADV ajst-10994 164 2 , , PUNCT ajst-10994 164 3 it it PRON ajst-10994 164 4 is be AUX ajst-10994 164 5 essential essential ADJ ajst-10994 164 6 to to PART ajst-10994 164 7 consider consider VERB ajst-10994 164 8 the the DET ajst-10994 164 9 model model NOUN ajst-10994 164 10 's 's PART ajst-10994 164 11 interpretability interpretability NOUN ajst-10994 164 12 and and CCONJ ajst-10994 164 13 computational computational ADJ ajst-10994 164 14 efficiency efficiency NOUN ajst-10994 164 15 to to PART ajst-10994 164 16 enhance enhance VERB ajst-10994 164 17 its its PRON ajst-10994 164 18 applicability applicability NOUN ajst-10994 164 19 and and CCONJ ajst-10994 164 20 acceptance acceptance NOUN ajst-10994 164 21 in in ADP ajst-10994 164 22 real real ADJ ajst-10994 164 23 - - PUNCT ajst-10994 164 24 world world NOUN ajst-10994 164 25 scenarios scenario NOUN ajst-10994 164 26 . . PUNCT ajst-10994 165 1 references reference NOUN ajst-10994 165 2 [ [ X ajst-10994 165 3 1 1 NUM ajst-10994 165 4 ] ] PUNCT ajst-10994 165 5 x. x. PROPN ajst-10994 165 6 zhang zhang PROPN ajst-10994 165 7 , , PUNCT ajst-10994 165 8 y. y. PROPN ajst-10994 165 9 liang liang PROPN ajst-10994 165 10 , , PUNCT ajst-10994 165 11 and and CCONJ ajst-10994 165 12 j. j. PROPN ajst-10994 165 13 zhou zhou PROPN ajst-10994 165 14 , , PUNCT ajst-10994 165 15 et et PROPN ajst-10994 165 16 al al PROPN ajst-10994 165 17 , , PUNCT ajst-10994 165 18 โ€œ " PUNCT ajst-10994 165 19 a a DET ajst-10994 165 20 novel novel NOUN ajst-10994 165 21 bearing bear VERB ajst-10994 165 22 fault fault NOUN ajst-10994 165 23 diagnosis diagnosis NOUN ajst-10994 165 24 model model NOUN ajst-10994 165 25 integrated integrate VERB ajst-10994 165 26 permutation permutation NOUN ajst-10994 165 27 entropy entropy NOUN ajst-10994 165 28 , , PUNCT ajst-10994 165 29 ensemble ensemble ADJ ajst-10994 165 30 empirical empirical ADJ ajst-10994 165 31 mode mode NOUN ajst-10994 165 32 decomposition decomposition NOUN ajst-10994 165 33 and and CCONJ ajst-10994 165 34 optimized optimize VERB ajst-10994 165 35 svm svm NOUN ajst-10994 165 36 , , PUNCT ajst-10994 165 37 โ€ " PUNCT ajst-10994 165 38 vol vol NOUN ajst-10994 165 39 . . PROPN ajst-10994 166 1 69 69 NUM ajst-10994 166 2 , , PUNCT ajst-10994 166 3 no no INTJ ajst-10994 166 4 . . NOUN ajst-10994 166 5 30 30 NUM ajst-10994 166 6 , , PUNCT ajst-10994 166 7 pp pp ADJ ajst-10994 166 8 . . PUNCT ajst-10994 166 9 164 164 NUM ajst-10994 166 10 - - SYM ajst-10994 166 11 179 179 NUM ajst-10994 166 12 , , PUNCT ajst-10994 166 13 2021 2021 NUM ajst-10994 166 14 . . PUNCT ajst-10994 167 1 [ [ X ajst-10994 167 2 2 2 NUM ajst-10994 167 3 ] ] X ajst-10994 167 4 levent levent NOUN ajst-10994 167 5 e e PROPN ajst-10994 167 6 , , PUNCT ajst-10994 167 7 turker turker NOUN ajst-10994 167 8 i i PRON ajst-10994 167 9 , , PUNCT ajst-10994 167 10 and and CCONJ ajst-10994 167 11 serkan serkan PROPN ajst-10994 167 12 k k PROPN ajst-10994 167 13 , , PUNCT ajst-10994 167 14 โ€œ " PUNCT ajst-10994 167 15 a a DET ajst-10994 167 16 generic generic ADJ ajst-10994 167 17 intelligent intelligent ADJ ajst-10994 167 18 bearing bearing NOUN ajst-10994 167 19 fault fault NOUN ajst-10994 167 20 diagnosis diagnosis NOUN ajst-10994 167 21 system system NOUN ajst-10994 167 22 using use VERB ajst-10994 167 23 compact compact ADJ ajst-10994 167 24 adaptive adaptive ADJ ajst-10994 167 25 1d 1d NUM ajst-10994 167 26 cnn cnn NOUN ajst-10994 167 27 classifier classifier NOUN ajst-10994 167 28 , , PUNCT ajst-10994 167 29 โ€ " PUNCT ajst-10994 167 30 journal journal NOUN ajst-10994 167 31 of of ADP ajst-10994 167 32 signal signal NOUN ajst-10994 167 33 processing processing NOUN ajst-10994 167 34 systems system NOUN ajst-10994 167 35 , , PUNCT ajst-10994 167 36 vol vol NOUN ajst-10994 167 37 . . PROPN ajst-10994 167 38 91 91 NUM ajst-10994 167 39 , , PUNCT ajst-10994 167 40 no no INTJ ajst-10994 167 41 . . NOUN ajst-10994 167 42 20 20 NUM ajst-10994 167 43 , , PUNCT ajst-10994 167 44 pp pp ADJ ajst-10994 167 45 . . PUNCT ajst-10994 168 1 179 179 NUM ajst-10994 168 2 - - SYM ajst-10994 168 3 189 189 NUM ajst-10994 168 4 , , PUNCT ajst-10994 168 5 2019 2019 NUM ajst-10994 168 6 . . PUNCT ajst-10994 169 1 [ [ X ajst-10994 169 2 3 3 X ajst-10994 169 3 ] ] X ajst-10994 169 4 h. h. PROPN ajst-10994 169 5 fang fang PROPN ajst-10994 169 6 , , PUNCT ajst-10994 169 7 j. j. PROPN ajst-10994 169 8 deng deng PROPN ajst-10994 169 9 , , PUNCT ajst-10994 169 10 and and CCONJ ajst-10994 169 11 y. y. PROPN ajst-10994 169 12 bai bai PROPN ajst-10994 169 13 , , PUNCT ajst-10994 169 14 et et PROPN ajst-10994 169 15 al al PROPN ajst-10994 169 16 , , PUNCT ajst-10994 169 17 โ€œ " PUNCT ajst-10994 169 18 clformer clformer NOUN ajst-10994 169 19 : : PUNCT ajst-10994 169 20 a a DET ajst-10994 169 21 lightweight lightweight ADJ ajst-10994 169 22 transformer transformer NOUN ajst-10994 169 23 based base VERB ajst-10994 169 24 on on ADP ajst-10994 169 25 convolutional convolutional ADJ ajst-10994 169 26 embedding embedding NOUN ajst-10994 169 27 and and CCONJ ajst-10994 169 28 linear linear ADJ ajst-10994 169 29 selfattention selfattention NOUN ajst-10994 169 30 with with ADP ajst-10994 169 31 strong strong ADJ ajst-10994 169 32 robustness robustness NOUN ajst-10994 169 33 for for ADP ajst-10994 169 34 fault fault NOUN ajst-10994 169 35 diagnosis diagnosis NOUN ajst-10994 169 36 under under ADP ajst-10994 169 37 limited limited ADJ ajst-10994 169 38 sample sample NOUN ajst-10994 169 39 conditions condition NOUN ajst-10994 169 40 , , PUNCT ajst-10994 169 41 โ€ " PUNCT ajst-10994 169 42 ieee ieee NOUN ajst-10994 169 43 transactions transaction NOUN ajst-10994 169 44 on on ADP ajst-10994 169 45 instrumentation instrumentation NOUN ajst-10994 169 46 and and CCONJ ajst-10994 169 47 measurement measurement NOUN ajst-10994 169 48 , , PUNCT ajst-10994 169 49 no no INTJ ajst-10994 169 50 . . NOUN ajst-10994 169 51 71 71 NUM ajst-10994 169 52 , , PUNCT ajst-10994 169 53 pp pp ADJ ajst-10994 169 54 . . PUNCT ajst-10994 170 1 1 1 NUM ajst-10994 170 2 - - SYM ajst-10994 170 3 8 8 NUM ajst-10994 170 4 , , PUNCT ajst-10994 170 5 2022 2022 NUM ajst-10994 170 6 . . PUNCT ajst-10994 171 1 [ [ X ajst-10994 171 2 4 4 X ajst-10994 171 3 ] ] PUNCT ajst-10994 171 4 s. s. PROPN ajst-10994 171 5 zhou zhou PROPN ajst-10994 171 6 , , PUNCT ajst-10994 171 7 l. l. PROPN ajst-10994 171 8 wu wu PROPN ajst-10994 171 9 , , PUNCT ajst-10994 171 10 and and CCONJ ajst-10994 171 11 h. h. PROPN ajst-10994 171 12 su su PROPN ajst-10994 171 13 , , PUNCT ajst-10994 171 14 โ€œ " PUNCT ajst-10994 171 15 random random ADJ ajst-10994 171 16 exponential exponential ADJ ajst-10994 171 17 attractors attractor NOUN ajst-10994 171 18 for for ADP ajst-10994 171 19 a a DET ajst-10994 171 20 non non ADJ ajst-10994 171 21 - - ADJ ajst-10994 171 22 autonomous autonomous ADJ ajst-10994 171 23 fitzhugh fitzhugh PROPN ajst-10994 171 24 - - PUNCT ajst-10994 171 25 nagumo nagumo ADJ ajst-10994 171 26 lattice lattice NOUN ajst-10994 171 27 system system NOUN ajst-10994 171 28 with with ADP ajst-10994 171 29 multiplicable multiplicable NOUN ajst-10994 171 30 white white ADJ ajst-10994 171 31 noise noise NOUN ajst-10994 171 32 , , PUNCT ajst-10994 171 33 โ€ " PUNCT ajst-10994 171 34 journal journal NOUN ajst-10994 171 35 of of ADP ajst-10994 171 36 zhejiang zhejiang PROPN ajst-10994 171 37 normal normal PROPN ajst-10994 171 38 university university PROPN ajst-10994 171 39 , , PUNCT ajst-10994 171 40 vol vol NOUN ajst-10994 171 41 . . PROPN ajst-10994 171 42 42 42 NUM ajst-10994 171 43 , , PUNCT ajst-10994 171 44 no no INTJ ajst-10994 171 45 . . NOUN ajst-10994 171 46 1 1 NUM ajst-10994 171 47 , , PUNCT ajst-10994 171 48 pp pp ADJ ajst-10994 171 49 . . PUNCT ajst-10994 172 1 1 1 NUM ajst-10994 172 2 - - SYM ajst-10994 172 3 8 8 NUM ajst-10994 172 4 , , PUNCT ajst-10994 172 5 2019 2019 NUM ajst-10994 172 6 . . PUNCT ajst-10994 173 1 [ [ X ajst-10994 173 2 5 5 X ajst-10994 173 3 ] ] PUNCT ajst-10994 173 4 h. h. PROPN ajst-10994 173 5 zhang zhang PROPN ajst-10994 173 6 , , PUNCT ajst-10994 173 7 x. x. PROPN ajst-10994 173 8 wang wang PROPN ajst-10994 173 9 , , PUNCT ajst-10994 173 10 and and CCONJ ajst-10994 173 11 x. x. PROPN ajst-10994 173 12 li li PROPN ajst-10994 173 13 , , PUNCT ajst-10994 173 14 โ€œ " PUNCT ajst-10994 173 15 low low ADJ ajst-10994 173 16 signal signal NOUN ajst-10994 173 17 - - PUNCT ajst-10994 173 18 to to ADP ajst-10994 173 19 - - PUNCT ajst-10994 173 20 noise noise NOUN ajst-10994 173 21 ratio ratio NOUN ajst-10994 173 22 noise noise NOUN ajst-10994 173 23 reduction reduction NOUN ajst-10994 173 24 algorithm algorithm NOUN ajst-10994 173 25 based base VERB ajst-10994 173 26 on on ADP ajst-10994 173 27 adaptive adaptive ADJ ajst-10994 173 28 threshold threshold NOUN ajst-10994 173 29 active active ADJ ajst-10994 173 30 speech speech NOUN ajst-10994 173 31 detection detection NOUN ajst-10994 173 32 and and CCONJ ajst-10994 173 33 minimum minimum NOUN ajst-10994 173 34 mean mean NOUN ajst-10994 173 35 square square NOUN ajst-10994 173 36 error error NOUN ajst-10994 173 37 log log NOUN ajst-10994 173 38 - - PUNCT ajst-10994 173 39 spectral spectral ADJ ajst-10994 173 40 amplitude amplitude NOUN ajst-10994 173 41 estimation estimation NOUN ajst-10994 173 42 , , PUNCT ajst-10994 173 43 โ€ " PUNCT ajst-10994 173 44 journal journal NOUN ajst-10994 173 45 of of ADP ajst-10994 173 46 research research NOUN ajst-10994 173 47 in in ADP ajst-10994 173 48 science science NOUN ajst-10994 173 49 and and CCONJ ajst-10994 173 50 engineering engineering NOUN ajst-10994 173 51 , , PUNCT ajst-10994 173 52 vol vol NOUN ajst-10994 173 53 . . PROPN ajst-10994 173 54 40 40 NUM ajst-10994 173 55 , , PUNCT ajst-10994 173 56 no no INTJ ajst-10994 173 57 . . NOUN ajst-10994 173 58 6 6 NUM ajst-10994 173 59 , , PUNCT ajst-10994 173 60 pp pp ADJ ajst-10994 173 61 . . PUNCT ajst-10994 174 1 1763 1763 NUM ajst-10994 174 2 - - SYM ajst-10994 174 3 1768 1768 NUM ajst-10994 174 4 , , PUNCT ajst-10994 174 5 2019 2019 NUM ajst-10994 174 6 . . PUNCT ajst-10994 175 1 [ [ X ajst-10994 175 2 6 6 NUM ajst-10994 175 3 ] ] PUNCT ajst-10994 175 4 x. x. PROPN ajst-10994 175 5 ma ma PROPN ajst-10994 175 6 , , PUNCT ajst-10994 175 7 j. j. PROPN ajst-10994 175 8 yu yu PROPN ajst-10994 175 9 , , PUNCT ajst-10994 175 10 and and CCONJ ajst-10994 175 11 f. f. PROPN ajst-10994 175 12 yang yang PROPN ajst-10994 175 13 , , PUNCT ajst-10994 175 14 et et PROPN ajst-10994 175 15 al al PROPN ajst-10994 175 16 , , PUNCT ajst-10994 175 17 โ€œ " PUNCT ajst-10994 175 18 pseudo pseudo ADJ ajst-10994 175 19 - - ADJ ajst-10994 175 20 random random ADJ ajst-10994 175 21 sequence sequence NOUN ajst-10994 175 22 generator generator NOUN ajst-10994 175 23 based base VERB ajst-10994 175 24 on on ADP ajst-10994 175 25 high high ADJ ajst-10994 175 26 dimensional dimensional ADJ ajst-10994 175 27 chaotic chaotic ADJ ajst-10994 175 28 system system NOUN ajst-10994 175 29 , , PUNCT ajst-10994 175 30 โ€ " PUNCT ajst-10994 175 31 journal journal NOUN ajst-10994 175 32 of of ADP ajst-10994 175 33 dalian dalian PROPN ajst-10994 175 34 institute institute PROPN ajst-10994 175 35 of of ADP ajst-10994 175 36 light light PROPN ajst-10994 175 37 industry industry NOUN ajst-10994 175 38 , , PUNCT ajst-10994 175 39 vol vol NOUN ajst-10994 175 40 . . PROPN ajst-10994 175 41 39 39 NUM ajst-10994 175 42 , , PUNCT ajst-10994 175 43 no no INTJ ajst-10994 175 44 . . NOUN ajst-10994 175 45 2 2 NUM ajst-10994 175 46 , , PUNCT ajst-10994 175 47 pp pp ADJ ajst-10994 175 48 . . PUNCT ajst-10994 176 1 143 143 NUM ajst-10994 176 2 - - SYM ajst-10994 176 3 149 149 NUM ajst-10994 176 4 , , PUNCT ajst-10994 176 5 2020 2020 NUM ajst-10994 176 6 . . PUNCT ajst-10994 177 1 [ [ X ajst-10994 177 2 7 7 X ajst-10994 177 3 ] ] PUNCT ajst-10994 177 4 z. z. PROPN ajst-10994 177 5 jiang jiang PROPN ajst-10994 177 6 , , PUNCT ajst-10994 177 7 and and CCONJ ajst-10994 177 8 q. q. PROPN ajst-10994 177 9 yang yang PROPN ajst-10994 177 10 , , PUNCT ajst-10994 177 11 โ€œ " PUNCT ajst-10994 177 12 generalizedcodeindexmodulationbasedondirectsequencespreadspectrum generalizedcodeindexmodulationbasedondirectsequencespreadspectrum NOUN ajst-10994 177 13 , , PUNCT ajst-10994 177 14 โ€ " PUNCT ajst-10994 177 15 application application NOUN ajst-10994 177 16 research research NOUN ajst-10994 177 17 of of ADP ajst-10994 177 18 computers computer NOUN ajst-10994 177 19 , , PUNCT ajst-10994 177 20 vol vol NOUN ajst-10994 177 21 . . PROPN ajst-10994 178 1 36 36 NUM ajst-10994 178 2 , , PUNCT ajst-10994 178 3 no no INTJ ajst-10994 178 4 . . NOUN ajst-10994 178 5 4 4 NUM ajst-10994 178 6 , , PUNCT ajst-10994 178 7 pp pp ADJ ajst-10994 178 8 . . PUNCT ajst-10994 179 1 1186 1186 NUM ajst-10994 179 2 - - SYM ajst-10994 179 3 1188 1188 NUM ajst-10994 179 4 , , PUNCT ajst-10994 179 5 2019 2019 NUM ajst-10994 179 6 . . PUNCT ajst-10994 180 1 [ [ X ajst-10994 180 2 8 8 X ajst-10994 180 3 ] ] X ajst-10994 180 4 jafarinejad jafarinejad PROPN ajst-10994 180 5 f f PROPN ajst-10994 180 6 , , PUNCT ajst-10994 180 7 pouyan pouyan VERB ajst-10994 180 8 a a DET ajst-10994 180 9 a a NOUN ajst-10994 180 10 , , PUNCT ajst-10994 180 11 โ€œ " PUNCT ajst-10994 180 12 a a DET ajst-10994 180 13 modular modular ADJ ajst-10994 180 14 synthesis synthesis NOUN ajst-10994 180 15 approach approach NOUN ajst-10994 180 16 for for ADP ajst-10994 180 17 intelligent intelligent ADJ ajst-10994 180 18 manufacturing manufacturing NOUN ajst-10994 180 19 system system NOUN ajst-10994 180 20 design design NOUN ajst-10994 180 21 : : PUNCT ajst-10994 180 22 a a DET ajst-10994 180 23 petri petri ADJ ajst-10994 180 24 net net ADJ ajst-10994 180 25 based base VERB ajst-10994 180 26 transformation transformation NOUN ajst-10994 180 27 method method NOUN ajst-10994 180 28 , , PUNCT ajst-10994 180 29 โ€ " PUNCT ajst-10994 180 30 signal signal NOUN ajst-10994 180 31 processing processing NOUN ajst-10994 180 32 and and CCONJ ajst-10994 180 33 intelligent intelligent ADJ ajst-10994 180 34 systems system NOUN ajst-10994 180 35 conference conference NOUN ajst-10994 180 36 ( ( PUNCT ajst-10994 180 37 spis spis PROPN ajst-10994 180 38 ) ) PUNCT ajst-10994 180 39 , , PUNCT ajst-10994 180 40 vol vol NOUN ajst-10994 180 41 . . PROPN ajst-10994 180 42 33 33 NUM ajst-10994 180 43 , , PUNCT ajst-10994 180 44 no no INTJ ajst-10994 180 45 . . NOUN ajst-10994 180 46 5 5 NUM ajst-10994 180 47 , , PUNCT ajst-10994 180 48 pp pp ADJ ajst-10994 180 49 . . PUNCT ajst-10994 181 1 40 40 NUM ajst-10994 181 2 - - SYM ajst-10994 181 3 45 45 NUM ajst-10994 181 4 , , PUNCT ajst-10994 181 5 2016 2016 NUM ajst-10994 181 6 . . PUNCT ajst-10994 182 1 [ [ X ajst-10994 182 2 9 9 NUM ajst-10994 182 3 ] ] SYM ajst-10994 182 4 kuehn kuehn NOUN ajst-10994 182 5 i i PROPN ajst-10994 182 6 , , PUNCT ajst-10994 182 7 cordier cordier PROPN ajst-10994 182 8 j j PROPN ajst-10994 182 9 j j PROPN ajst-10994 182 10 , , PUNCT ajst-10994 182 11 and and CCONJ ajst-10994 182 12 baylard baylard PROPN ajst-10994 182 13 c c NOUN ajst-10994 182 14 , , PUNCT ajst-10994 182 15 et et PROPN ajst-10994 182 16 al al PROPN ajst-10994 182 17 , , PUNCT ajst-10994 182 18 โ€œ " PUNCT ajst-10994 182 19 management management NOUN ajst-10994 182 20 of of ADP ajst-10994 182 21 the the DET ajst-10994 182 22 iter iter NOUN ajst-10994 182 23 buildings building NOUN ajst-10994 182 24 configuration configuration NOUN ajst-10994 182 25 for for ADP ajst-10994 182 26 the the DET ajst-10994 182 27 construction construction NOUN ajst-10994 182 28 and and CCONJ ajst-10994 182 29 installation installation NOUN ajst-10994 182 30 phase phase NOUN ajst-10994 182 31 , , PUNCT ajst-10994 182 32 โ€ " PUNCT ajst-10994 182 33 ieee ieee NOUN ajst-10994 182 34 transactions transaction NOUN ajst-10994 182 35 on on ADP ajst-10994 182 36 plasma plasma NOUN ajst-10994 182 37 science science NOUN ajst-10994 182 38 , , PUNCT ajst-10994 182 39 vol vol NOUN ajst-10994 182 40 . . PROPN ajst-10994 182 41 46 46 NUM ajst-10994 182 42 , , PUNCT ajst-10994 182 43 no no INTJ ajst-10994 182 44 . . NOUN ajst-10994 182 45 1 1 NUM ajst-10994 182 46 , , PUNCT ajst-10994 182 47 pp pp ADJ ajst-10994 182 48 . . PUNCT ajst-10994 183 1 194 194 NUM ajst-10994 183 2 - - SYM ajst-10994 183 3 200 200 NUM ajst-10994 183 4 , , PUNCT ajst-10994 183 5 2018 2018 NUM ajst-10994 183 6 . . PUNCT ajst-10994 184 1 [ [ X ajst-10994 184 2 10 10 NUM ajst-10994 184 3 ] ] PUNCT ajst-10994 184 4 somes some NOUN ajst-10994 184 5 n n CCONJ ajst-10994 184 6 , , PUNCT ajst-10994 184 7 and and CCONJ ajst-10994 184 8 moore moore PROPN ajst-10994 184 9 j j PROPN ajst-10994 184 10 , , PUNCT ajst-10994 184 11 โ€œ " PUNCT ajst-10994 184 12 design design NOUN ajst-10994 184 13 and and CCONJ ajst-10994 184 14 construction construction NOUN ajst-10994 184 15 of of ADP ajst-10994 184 16 a a DET ajst-10994 184 17 regional regional ADJ ajst-10994 184 18 scale scale NOUN ajst-10994 184 19 bioret bioret NOUN ajst-10994 184 20 ention ention NOUN ajst-10994 184 21 system system NOUN ajst-10994 184 22 , , PUNCT ajst-10994 184 23 โ€ " PUNCT ajst-10994 184 24 rainwater&urban rainwater&urban NUM ajst-10994 184 25 design design NOUN ajst-10994 184 26 , , PUNCT ajst-10994 184 27 vol vol NOUN ajst-10994 184 28 . . PROPN ajst-10994 184 29 20 20 NUM ajst-10994 184 30 , , PUNCT ajst-10994 184 31 no no INTJ ajst-10994 184 32 . . NOUN ajst-10994 184 33 3 3 NUM ajst-10994 184 34 , , PUNCT ajst-10994 184 35 pp pp ADJ ajst-10994 184 36 . . PUNCT ajst-10994 185 1 12 12 NUM ajst-10994 185 2 - - SYM ajst-10994 185 3 13 13 NUM ajst-10994 185 4 , , PUNCT ajst-10994 185 5 2007 2007 NUM ajst-10994 185 6 . . PUNCT ajst-10994 186 1 [ [ X ajst-10994 186 2 11 11 NUM ajst-10994 186 3 ] ] X ajst-10994 186 4 urban urban ADJ ajst-10994 186 5 k k PROPN ajst-10994 186 6 l l PROPN ajst-10994 186 7 , , PUNCT ajst-10994 186 8 scheller scheller NOUN ajst-10994 186 9 f f PROPN ajst-10994 186 10 , , PUNCT ajst-10994 186 11 and and CCONJ ajst-10994 186 12 bruckner bruckner PROPN ajst-10994 186 13 t t PROPN ajst-10994 186 14 , , PUNCT ajst-10994 186 15 โ€œ " PUNCT ajst-10994 186 16 suitability suitability NOUN ajst-10994 186 17 assessment assessment NOUN ajst-10994 186 18 of of ADP ajst-10994 186 19 models model NOUN ajst-10994 186 20 in in ADP ajst-10994 186 21 the the DET ajst-10994 186 22 industrial industrial ADJ ajst-10994 186 23 energy energy NOUN ajst-10994 186 24 system system NOUN ajst-10994 186 25 design design NOUN ajst-10994 186 26 , , PUNCT ajst-10994 186 27 โ€ " PUNCT ajst-10994 186 28 renewable renewable ADJ ajst-10994 186 29 and and CCONJ ajst-10994 186 30 sustainable sustainable ADJ ajst-10994 186 31 energy energy NOUN ajst-10994 186 32 reviews review NOUN ajst-10994 186 33 , , PUNCT ajst-10994 186 34 vol vol NOUN ajst-10994 186 35 . . PROPN ajst-10994 186 36 137 137 NUM ajst-10994 186 37 , , PUNCT ajst-10994 186 38 pp pp ADJ ajst-10994 186 39 . . PUNCT ajst-10994 187 1 110400 110400 NUM ajst-10994 187 2 - - SYM ajst-10994 187 3 110402 110402 NUM ajst-10994 187 4 , , PUNCT ajst-10994 187 5 2021 2021 NUM ajst-10994 187 6 . . PUNCT ajst-10994 188 1 [ [ X ajst-10994 188 2 12 12 NUM ajst-10994 188 3 ] ] X ajst-10994 188 4 kamrowska kamrowska ADJ ajst-10994 188 5 - - PUNCT ajst-10994 188 6 zauska zauska NOUN ajst-10994 188 7 d d PROPN ajst-10994 188 8 , , PUNCT ajst-10994 188 9 โ€œ " PUNCT ajst-10994 188 10 impact impact NOUN ajst-10994 188 11 of of ADP ajst-10994 188 12 ai ai NOUN ajst-10994 188 13 - - PUNCT ajst-10994 188 14 based base VERB ajst-10994 188 15 tools tool NOUN ajst-10994 188 16 and and CCONJ ajst-10994 188 17 urban urban ADJ ajst-10994 188 18 big big ADJ ajst-10994 188 19 data datum NOUN ajst-10994 188 20 analytics analytic NOUN ajst-10994 188 21 on on ADP ajst-10994 188 22 the the DET ajst-10994 188 23 design design NOUN ajst-10994 188 24 and and CCONJ ajst-10994 188 25 planning planning NOUN ajst-10994 188 26 of of ADP ajst-10994 188 27 cities city NOUN ajst-10994 188 28 , , PUNCT ajst-10994 188 29 โ€ " PUNCT ajst-10994 188 30 land land NOUN ajst-10994 188 31 , , PUNCT ajst-10994 188 32 vol vol NOUN ajst-10994 188 33 . . PROPN ajst-10994 188 34 10 10 NUM ajst-10994 188 35 , , PUNCT ajst-10994 188 36 pp pp ADJ ajst-10994 188 37 . . PUNCT ajst-10994 189 1 101 101 NUM ajst-10994 189 2 - - SYM ajst-10994 189 3 103 103 NUM ajst-10994 189 4 , , PUNCT ajst-10994 189 5 2021 2021 NUM ajst-10994 189 6 . . PUNCT ajst-10994 190 1 [ [ X ajst-10994 190 2 13 13 NUM ajst-10994 190 3 ] ] PUNCT ajst-10994 190 4 l. l. PROPN ajst-10994 190 5 deng deng PROPN ajst-10994 190 6 , , PUNCT ajst-10994 190 7 โ€œ " PUNCT ajst-10994 190 8 analysis analysis NOUN ajst-10994 190 9 on on ADP ajst-10994 190 10 the the DET ajst-10994 190 11 application application NOUN ajst-10994 190 12 and and CCONJ ajst-10994 190 13 strategy strategy NOUN ajst-10994 190 14 of of ADP ajst-10994 190 15 sponge sponge NOUN ajst-10994 190 16 city city NOUN ajst-10994 190 17 concept concept NOUN ajst-10994 190 18 in in ADP ajst-10994 190 19 landscape landscape NOUN ajst-10994 190 20 architecture architecture NOUN ajst-10994 190 21 planning planning NOUN ajst-10994 190 22 , , PUNCT ajst-10994 190 23 โ€ " PUNCT ajst-10994 190 24 smart smart ADJ ajst-10994 190 25 cities city NOUN ajst-10994 190 26 , , PUNCT ajst-10994 190 27 vol vol NOUN ajst-10994 190 28 . . PROPN ajst-10994 190 29 6 6 NUM ajst-10994 190 30 , , PUNCT ajst-10994 190 31 no no INTJ ajst-10994 190 32 . . NOUN ajst-10994 190 33 5 5 NUM ajst-10994 190 34 , , PUNCT ajst-10994 190 35 pp pp ADJ ajst-10994 190 36 . . PUNCT ajst-10994 191 1 143 143 NUM ajst-10994 191 2 - - SYM ajst-10994 191 3 144 144 NUM ajst-10994 191 4 , , PUNCT ajst-10994 191 5 2020 2020 NUM ajst-10994 191 6 . . PUNCT