Academic Journal of Science and Technology ISSN: 2771-3032 | Vol. 5, No. 1, 2023 95 A Review of the Research of Quick Access Recorder data Heng Zhang School of Civil Aviation Safety Engineering, Civil Aviation Flight University of China, Guanghan 618307, China Abstract: In recent years, QAR data has been widely concerned by scholars because of its reliability and integrity. QAR data has become an important basis for flight quality monitoring, engine status detection, aircraft system failure diagnosis, 3D animation route design and other aspects of various airlines in the world. The keywords of research on QAR data in China and other countries were clustering analyzed by VOSviewer, the hot spots were introduced, and research were summarized and discussed from five aspects. At last, some shortcomings of current research on QAR data were pointed out and some future development directions were presented. Keywords: QAR data, Aircraft engine, Fuel. 1. Introduction In 1997, the Civil Aviation Administration of China (CAAC) required all aircraft in operation to be equipped with the Quick Access Recorder (QAR) in CCAR-121, which was used to obtain the flight data of the daily operation of the aircraft. QAR system has greatly improved the efficiency and accuracy of flight data acquisition compared with the traditional follower flight. Current general QAR system will record the data in the disc or PC card, after the flight landed, through 3G or 4G mobile signal, the data transmitted to the data of the ground treatment station, QAR system has small volume, large capacity, easy access and the characteristics of high reliability compared with black box. At present, QAR data has become an important basis for the flight quality monitoring, engine condition detection, aircraft system fault diagnosis, 3D animation route design and other aspects of the world's airlines. In January 2018, CAAC Flight Quality Monitoring (FOQA) station was officially launched. It is a large-scale, centralized QAR data analysis and application system that is currently monitoring over 3,800 commercial aircraft. It processes and analyzes more than 16,000 flights and more than 150GB of QAR data every day, with functions such as decoding the original QAR data, data extraction, correlation analysis and risk assessment. The system has significantly improved CAAC's ability to supervise flight operation quality and safety. According to the International Civil Aviation Organization (ICAO) 's Fifth Meeting of the Spectrum Review Working Group(SRWG/5) held in March 2021, China is the only country in the world that uses QAR data to conduct comprehensive monitoring and assessment of all domestic commercial transport fleets, flights and typical emergencies. This paper classifies, combs and summarizes domestic and foreign researches on QAR data, so as to find out research hotspots and future research directions based on QAR data. 2. Key Words Analysis of China and Other Countries’ Research In this section, the VOSviewer software is used to conduct keyword cluster analysis in Chinese’s and Web Of Science database’s literatures based on QAR data respectively, and the keyword atlas of Chinese and other ciuntires’ researches is obtained. Then, the keyword atlas is used to analyze the researched keywords. Figure 1. Keyword cluster analysis of domestic research literature 96 2.1. Keywords analysis of QAR data research in China In order to ensure the authority and reliability of the literature, the Chinese literature collected in this paper was retrieved from CNKI, and the search keywords were: "QAR", "quick access recorder", "Quick access Recorder", the search scope for journals, and the search results were read and screened, eliminate irrelevant articles, the final determination of the number of Chinese literature is 220. The keyword cluster analysis was carried out, and the screening frequency was no less than 5 times. A total of 34 keywords meeting the conditions were obtained, and the following distribution diagram was obtained. After removing the search keywords such as "QAR" and "QAR data", the top 10 in the frequency ranking were Aero-engine (aero engine) 33 times. flight safety (28 times), flight data (24 times), fault diagnosis (23 times), fuel flow (19 times), data mining (13 times), feature extraction (10 times), regression analysis (10 times), anomaly detection (8 times) and flight quality monitoring (6 times). Through keyword atlas analysis, it can be concluded that: 1) The research on aircraft components and fault diagnosis based on QAR data is the most concerned field of domestic scholars, among which the most academic articles have been published on the monitoring of engine and its components; 2) Research on flight safety based on QAR data is another hot spot, including flight quality monitoring, risk assessment, etc.; 3) QAR data mining research is also a hot topic, including the improvement of the algorithm and the method of extracting characteristic parameters in QAR data; 4) The number of research articles on aircraft emissions based on QAR data is not too large, but it is also a key research direction that cannot be ignored. 2.2. Keywords analysis of QAR data research based on Web of Science The literatures were retrieved from the Web Of Science(WOS), the search keywords were "QAR data", "Quick Access Recorder", the search scope was the core set of WOS since 1990, the result of manual screening of the articles, the final number of foreign literatures was 98. Also, the VOSviewer software was used to conduct keyword cluster analysis on the retrieval results, and the frequency of occurrence was set to be no less than 2. There were 34 keywords meeting the conditions, and the distribution diagram was as follows, and the retrieval keywords such as "qar data" and "quick access recorder" were deleted. Keywords that appeared most frequently were flight data 6 times, fuel consumption 4 times, deep learning 3 times, and pilots 2 times. Figure 2. Keyword cluster analysis of foreign research literature Through keyword atlas analysis, it can be concluded that: 1) Similar to domestic research, improving data analysis algorithm based on QAR data is the research focus of scholars; 2) Research on fuel efficiency based on QAR data is one of the hot spots of scholars; 3) Different from Chinese studies, the study of pilot safety operation evaluation based on QAR data is a hot topic in papers. 3. Classification and Sorting of QAR Data Related Studies Through the analysis of keywords, combined with the four elements of the comprehensive management of civil aviation safety, the research on QAR data at home and abroad is summarized into five aspects: 1. Research on airplanes; 2. Research on QAR data; 3 Research on pilots; 4.Research on environmental impact; 5. Research on management. The following is a review of these five aspects of research. 3.1. The research on airplanes The research directions of airplanes and its related parts based on QAR data include fault detection of engine and its components, monitoring of fuel consumption efficiency, fault monitoring of airborne equipment and monitoring of 97 aerodynamics of airplanes. The following are the researches in these four directions respectively combed and elaborated. 3.1.1. Fault detection of engine and its components The performance of civil aviation engine directly affects the performance, reliability and economy of aircraft, which is an important embodiment of a country's science and technology, industry and national defense strength. Among the researches based on QAR data, the amount of researches on engine are the most. Early researchers were mainly composed of technical staff of airlines. Limited by the decoding system and computer technology provided by airlines, they could only monitor some parameters of engines [1,2]. With the development of computer technology, researchers can use QAR data to discriminate and monitor the state of the engine in different flight stages [3-5]. Monitoring data types are becoming more comprehensive and monitoring accuracy is becoming higher, which is conducive to avoiding the occurrence of unsafe state of the engine. In addition to monitoring the engine, researchers began to conduct in-depth mining and analysis of QAR data: Cao Huiling from Civil Aviation University of China obtained the exhaust temperature (EGT) margin of CFM56-7B engine by converting relevant parameters of QAR data, and then established the recession model of the engine by segentially fitting [6]. Then, an engine anomaly detection system was developed by combining QAR data with support vector machine (SVR) algorithm [7], and an evaluation model of engine health state was established by combining the relevant parameters of QAR data with the analytic hierarchy process (AHP) and grey correlation analysis [8]. With the deepening of the research, other researchers began to study the aircraft power system based on the principle of thermodynamics: analyze the parameters related to the aircraft engine bleed air system and auxiliary power system from the QAR data, and carry out fault detection and prediction [9,10]. Based on QAR data, specific unit components of the engine are studied, such as the complex structure and regulation rules of turbine blade [11,12], rotor [13] and adjustable vent valve [14-16], and the prediction model is obtained through mathematical simulation. Continuous detailed research can enable airlines to make more accurate maintenance plans for the engine and its specific components, ensure the normal operation of the engine, and improve the safety management level of airlines.。 3.1.2. Monitoring of fuel efficiency The need to improve the efficiency of monitoring fuel consumption and control fuel costs has existed since the invention of airplanes. After QAR devices were widely installed on airplanes, researchers began to mine the parameters related to fuel efficiency from QAR data. In the initial research, mathematical statistics were used to model and analyze the fuel consumption of the engine during take- off [17], cruise, landing and ground taxi [18]. On this basis, Cao Huiling et al. [19] used multivariate statistics algorithm and MATLAB platform to establish a simulation platform applicable to various engine models to estimate fuel consumption of each segment. At the same time, with the development of big data theory and computer algorithms, the research of new algorithms based on correlation degree and signal decomposition began to appear. At the same time, the proposed and continuous optimization of neural network and machine learning algorithm [20,21], researchers applied the method of processing big data to the research of fuel consumption and the establishment of prediction model. The establishment of fuel consumption model can better estimate engine performance, control fuel saving flow, and make more accurate prediction of fuel consumption, but there are also time consuming and workload problems. Different from the above modeling and simulation analysis methods, Ye[22] et al. proposed a method based on the inherent mechanical characteristics of the engine without modeling. By using QAR data to match flight conditions, fuel flow prediction is carried out. This method greatly reduces the verification time and cost, and provides a new way to study QAR data. In addition to engine fuel, lubricating oil and other oils are also important to ensure aircraft flight safety. Wang et al. [23] built a model by mining parameters such as oil consumption rate in QAR data to judge the health of aeroengine lubrication system. 3.1.3. Airborne equipment failure monitoring In addition to the research on engine power system faults, Wu Renbiao et al. [24] developed a fault diagnosis simulation system for airborne equipment based on Simulink and Microsoft Visual Basic 6.0 software. Through monitoring various parameters of QAR data, they found fault symptoms. This enables maintenance personnel to perform troubleshooting, restricted operation, or replacement in a timely manner to ensure flight safety. Among all the studies on airborne equipment, the air conditioning system of aircraft is the key research object. By monitoring and fault detection of the flow of aircraft air conditioning components [25], researchers classify the working state of the air conditioning system and establish a degradation assessment algorithm based on the physical characteristics of the air conditioning system, so as to achieve the purpose of monitoring and evaluation of the air conditioning system of aircraft. To facilitate airlines to arrange flight schedules and maintenance schedules. 3.1.4. Monitoring of aerodynamics of aircraft The aerodynamic model of civil aircraft has been determined after its manufacturing. However, structural fatigue or corrosion caused by long-term flight will lead to minor changes in the aerodynamic model of the aircraft. Accurate acquisition of relevant parameters of the aerodynamic model from QAR data is conducive to safe flight operation of aircraft in different environments [26]. It has important practical significance such as flight simulation, maintenance information guidance and accident warning and recurrence. Through combing, it is found that the research of aircraft and its components based on QAR data has made some achievements. And the research develops from a certain system to a certain part of the system, the research object is more and more detailed; It is beneficial to establish the decay model of the system or component based on the physical characteristics of each system or component of the aircraft, which is conducive to the arrangement of flight planning and maintenance plan of airlines. Monitoring combined with fault detection model can effectively detect the fault of aircraft systems and components. The monitoring of specific aircraft components and performance attenuation prediction are the development trend of QAR data for aircraft research. 3.2. The research on QAR data QAR equipment records all data related to flight in the course of flight. The study of the data itself can dig deeper information in it, and improve the monitoring efficiency of 98 flight process and the utilization rate of QAR data of airlines. With the rapid development of the civil aviation industry, the quantity of QAR data is also increasing rapidly. Research on QAR data itself runs through the research on aircraft, management, environment and pilots based on QAR data. Meanwhile, with the development of computer technology, research methods on QAR data are also improving. At the early stage of the application of QAR equipment, airlines can only use the aircraft performance monitoring software provided by the aircraft manufacturer to carry out decoding, analysis and calculation of the collected data, and can only carry out aircraft performance monitoring based on the results. Due to the poor software analysis ability and the different analysis software used by different airlines, a large amount of QAR data is wasted. In order to solve this problem, one of the earliest ideas is to establish a database with higher storage efficiency. Researchers start from the perspective of reducing QAR data, respectively using compressed sensing theory, wavelet transform, cluster analysis and other methods to establish a new QAR data storage system, so as to store QAR data better and prevent data from being overwritten and wasted. Another idea is to improve the decoding speed of QAR data from the perspective of data processing speed, using dynamic linked list, the establishment of automatic decoding system and other methods, so as to improve the utilization of QAR data. At the same time, researchers analyzed the abnormal data in QAR data and carried out de- noising, detection and correction to improve the reliability of the data [27]. QAR data records all the data in the flight process. However, in the research process, only a certain class of problems are analyzed, so only parameters in the QAR data that are highly relevant to the analyzed problems need to be found for analysis. This step is called feature extraction of QAR data. Feature extraction is the most basic step in all current QAR data-based research, which can greatly simplify the amount of data analyzed. There are qualitative and quantitative methods to extract characteristic parameters. The qualitative method mainly refers to the selection of relevant parameters according to expert opinions. Quantitative method refers to the use of computer mathematical methods to extract the relevant parameters. Qualitative method and quantitative method should be unified and complement each other. The development of quantitative analysis methods is described below. The quantitative method of feature extraction is mainly based on association rule analysis, time series, regression model, neural network, ball vector machine, cluster analysis [28] and other mathematical methods for feature extraction of QAR data. With the continuous development of machine learning theory, researchers have begun to use deep confidence network, convolutional neural network combined with short and long term memory network, Transformer network and other technologies based on deep learning [29] to improve the feature extraction method of QAR data, so as to improve the utilization rate of QAR data. Emerging services such as cloud computing, the Internet of Things, and social networks have contributed to an unprecedented increase in the type and scale of data in human society, and the era of big data has officially arrived. Data has changed from a simple object to a basic resource. How to better manage and utilize data has become a topic of general concern. The scale effect of big data brings great challenges to data storage, management and analysis, and changes in data management methods are brewing and occurring. 3.3. The research of pilots The pilot is the pilot of the aircraft, and the operation of the pilot is closely related to flight safety. In civil aviation production activities, potential hazards and events can be referred to as over-limit events, and pilot over-limit behavior can be understood as potential pilot operations that may cause accidents. Research on pilots based on QAR data is often conducted indirectly according to parameters related to overrun events. Typical overrun events include excessive pitch Angle and sideslip Angle during takeoff and landing. If the over-limit event is not found and improved in time, heavy landing, tail rubbing and other unsafe events may occur [30,31], or even cause serious losses. By counting the occurrence frequency of QAR over-limit events as the evaluation standard, the safety risk assessment method is introduced to establish a model to analyze the QAR data, so as to evaluate, analyze and warn the flight operation of pilots. The pilots with poor evaluation results are focused on and the training requirements are put forward for them. Airlines can also take the evaluation results as one of the bases for year-end performance rewards and punishments [32]. At the same time, some researchers also proposed that psychology [33] and QAR data should be integrated to evaluate the operation of pilots, and the mental and psychological state of pilots should be fully considered. Different pilots have different ability to bear risks, and external pressure may also have an impact on the operation of pilots, so the comprehensive evaluation should be conducted. With the deepening of the research, a more comprehensive consideration of various factors in the work of pilots, so as to comprehensively evaluate the flight level of pilots will become the development direction of pilot research using QAR data in the future. 3.4. The research of environmental impact As aviation has grown rapidly, so have emissions from aircraft. With the enhancement of global environmental awareness and the vigorous implementation of the green development of civil aviation in China, the issue of aircraft exhaust emission has gradually attracted the attention of researchers, and the study of aircraft emission pollution has become a hot topic based on QAR data. The pollutants emitted by aircraft engines mainly include gas pollutants and particulate pollutants. Since the calculation of particle pollutants is more complicated than that of gas, the initial research focused on gas pollutants. Researchers obtained engine emission index and emissions by using parameters such as temperature, humidity, altitude, speed and fuel flow in QAR data, evaluated aircraft diffusion based on the results [34], and provided a basis for its emission control measures. With the construction of airport economic zones in recent years and the public's extensive attention to air quality, researchers began to monitor and evaluate aircraft emissions in the airport area by combining engine performance data with airport atmospheric environment parameters [35] [36]. With the development of technology and the attention paid to the impact of particle pollutants, researches on particle pollutants began to emerge in large numbers. Cao Huiling [37] et al. summarized the researches on gas pollutants and evaluated the black carbon emission characteristics of the whole flight segment of an aircraft considering the performance differences of engines, combustion quality and external environmental conditions. It provides more accurate 99 basis for aircraft emission monitoring and equivalent replacement of airworthiness compliance verification. The research on aircraft emissions based on QAR data can propose emission reduction methods for airlines from a scientific perspective, such as improving fleet mix, optimizing scene operation procedures, adjusting route and cruise mode, etc. Different researchers pay attention to the emissions or emissions in different flight stages. All researchers are trying to obtain more accurate emissions, but there is still no model that can accurately calculate the accurate emissions of different pollutants in different flight stages. With the deepening of the research on aircraft exhaust emissions, the emissions of both gas and particulate pollutants will become more and more accurate, and the impact will be more predictable, enabling airlines to make more accurate plans for energy conservation and emission reduction. 3.5. The research of management The use of QAR data can improve the management efficiency of airlines: Researchers use high-precision synthesis and integration, cluster analysis methods to analyze QAR data and flight route data; Airlines can find existing problems based on abnormal data, arrange flight plans and optimize routes [38], reduce flight delays [39], and improve the benefits of airlines. In addition to improving efficiency, relevant managers can obtain flight quality-related parameters such as flight track from QAR data and evaluate flight safety at different flight stages by combining risk assessment model and other methods [40,41]. Airlines can timely discover flights and routes with poor flight quality and propose targeted improvement measures. Data mining and machine learning methods are adopted to evaluate flight quality, which greatly improves the accuracy of flight quality assessment and is conducive to improving flight quality and flight safety [42]. In addition, some research results of the above mentioned studies on aircraft, QAR data and management work will also have certain reference significance for management work, such as the research on the designation of maintenance plan in aircraft research, which can also improve the efficiency of management work to a certain extent. In the future, studies on management using QAR data, whether safety management or efficiency management, will integrate the above mentioned studies on aircraft, QAR data, pilots and environment to comprehensively improve the management ability of management. 4. Problems and Future Research Directions Through the above combing, some problems and development directions of QAR data research at home and abroad can be summarized, including: 1) Research using QAR data is becoming more and more comprehensive. Among them, monitoring the health of various system components of aircraft and monitoring the flight quality are the main research areas at present. At the same time, with the development of theories related to human factors and green development, researches on pilot pressure and environmental protection have emerged in large numbers, which have become the hot spot of QAR data research in recent years. With the deepening of the research, further mining of the parameter information of QAR data to specifically study the work and attenuation characteristics of a certain part of an aircraft, or pilot operation and aircraft exhaust emissions will become the development direction of QAR data research in the future. 2) Research on QAR data requires a relatively professional and extensive knowledge reserve. No matter the research of aircraft, QAR data, pilots, environment or management, there is a problem that the accuracy and speed of data processing by mathematical statistics can not meet the requirements of the increasing amount of data. At present, the method of machine learning is widely used for data analysis. While the algorithms used in data analysis continue to improve, there is a trend of multidisciplinary integration of research, which requires a high knowledge reserve of researchers. For example, in the study of pilots, they should not only be familiar with the relevant operations and parameters of aircraft flight, but also master the data processing methods to process and analyze QAR data. Further research also requires researchers to master the relevant knowledge of psychology and risk assessment. Interdisciplinary knowledge reserve has become a necessary quality requirement for researchers. 3) There is still room for further mining of QAR data. QAR data parameters are numerous, different research objects need to study different parameters, and the relationship between parameters has not been fully studied at present, it can be combined with relevant theories and algorithms in data mining, applied to the mining of QAR data parameter association. The second chapter of this paper classifies the researches based on QAR data, but in fact, each category does not exist independently, but there are potential connections, and these connections need to be further explored by researchers. With the deepening of the research, two or more aspects of the above five aspects should be combined into the research. For example, the research on aircraft exhaust emissions mentioned in 2.5 can be applied to the management work and can be a direction of future research. 4) Over-emphasis on QAR data exists. Since our civil aviation regulatory authorities begin to use QAR data to analyze and evaluate the flight quality of flights, airlines have begun to attach importance to the QAR data and link it with the pilot's performance. As a result, pilots have many flight operations to ensure the good appearance of QAR data, which may lead to operation conflicts and the possibility of unsafe incidents. This requires airlines to pay reasonable attention to QAR data, and should not only rely on QAR data to evaluate pilot operations. In the future, more parameters should be added for comprehensive evaluation of flight quality. 5. Closing Remarks With the official arrival of the era of big data, QAR data has changed from a simple processing object to a basic resource for airlines and Civil Aviation Administration. How to better manage and use QAR data has become a topic of general concern. Making full use of the existing value can effectively improve the efficiency of airlines and enhance the work efficiency of regulatory authorities. In this paper, the research results of QAR data are introduced. In order to facilitate sorting, five categories of aircraft, QAR data, pilots, environment and management work are sorted respectively. In addition, this paper also points out the problems existing in the current QAR data and points out the development direction of future research. 100 References [1] CAO Huiling, ZHOU Baizheng. Application Research of QAR Data on Aero-Engine Monitoring [J]. Journal of Civil Aviation University of China, 2010, 28(03): 15-19. [2] SHI Xiu-yu. Fault Diagnosis Approach of Performance for Civil Aeroengine[J]. Aeroengine, 2008(03): 49-51. [3] WANG Yiwei, MO Liping, WANG Yishou,et al. Aero-engine status identification based on full-segment QAR data and convolutional nerural network[J]. Journal of Aerospace Power, 2021, 36(07): 1556-1563. [4] WANG Yi-shou, YU Ying-hong, QING Xin-lin, et al. Exhaust Gas Temperature Baseline Model of Aeroengine Based on Kernel Principal Component Analysis [J]. Aeroengine, 2020, 46(01): 54-60. [5] SHI Hongwei. Implementation and development of fault monitoring by airlines using ACMS and QAR data [J]. Aviation Maintenance & Engineering, 2021(02): 18-20. [6] CAO Huiling, LIN Dajin, ZENG Xuefeng. Declining Formula Establish and Analysis of CFM56-7B Aero-Engine [J]. Journal of Civil Aviation University of China, 2010, 28(04): 9-12. [7] Cao Huiling, Yang Lu, Lin Yusen, et al. Aero-engine Anomaly Detection Using Support Vector Ｒegression [J]. Mechanical Science and Technology for Aerospace Engineering, 2013, 32(11): 1616-1619. [8] CAO Huiling,HUANG Leteng,KANG Liping. Research on engine health assessment based on grey correlation analysis method [J]. Mathematics In Practice And Theory, 2015, 45(02): 122-129. [9] Gao X L, Zuo H F, Sun J Z, et al. Civil Aircraft Engine Start System Health Monitoring Method Based on QAR Data[M]. New York: Ieee, 2017: 168-173. [10] Xu Y J, Hou W K, Li W Z, et al.: Aero-Engine Gas-path Fault Diagnosis Based on Spatial Structural Characteristics of QAR Data, 2018 Annual Reliability and Maintainability Symposium, New York: Ieee, 2018. [11] CAO Huiling, ZHANG Hao. Fatigue Life Prediction of CFM56-7B Engine High Pressure Turbine Blades [J]. Aviation Maintenance & Engineering, 2021(07): 89-91. [12] HUANG Lei, GAO Shuwei. Load Spectrum Prediction of Engine High Pressure Turbine Blade Based on Numerical Simulation [J]. Electronic Technology, 2020, 49(03): 120-121. [13] Liu J Q, Feng Y W, Lu C, et al. Vibration Reliability Analysis of Aeroengine Rotor Based on Intelligent Neural Network Modeling Framework[J]. Shock and Vibration, 2021, 2021: 11. [14] TAN Yan, WEI Wuguo. Research on Working Baseline of Adjustable Discharge Valve of Aeroengine Based on Support Vector Regression [J]. Mathematics In Practice And Theory, 2020, 50(12): 22-27. [15] FENG Xiao, ZHANG Enyi. Fault Analysis and Status Monitoring of LEAP-1A Engine Starting Valve Based on QAR Data [J]. Aviation Maintenance & Engineering, 2021(05): 57- 59. [16] Tan Y, Iop. Fitting Operation Curve of Civil Aviation Turbo- fan Engine's Variable Bleed Valve based on MATLAB[C]. International Seminar on Computer Science and Engineering Technology (SCSET), 2018. [17] Zhang M, Huang Q W, Liu S H, et al. Fuel Consumption Model of the Climbing Phase of Departure Aircraft Based on Flight Data Analysis[J]. Sustainability, 2019, 11(16): 23. [18] Oh E M, Kim H, Jeon D, et al.: A Model for Estimation of Fuel Consumption during Aircraft Taxi Operations, 2018 Ieee/Aiaa 37th Digital Avionics Systems Conference, New York: Ieee, 2018: 266-271. [19] CAO Huiling, JIA Chao. Research on fuel flow control law of civil aviation engine based on QAR [J]. Science Technology and Engineering, 2013, 13(13): 3814-3817+3827. [20] Wang K, Chen J J, Ieee. An Interval Prediction Method for Imbalanced Fuel Consumption Data[C]. Chinese Automation Congress (CAC), 2020: 824-829. [21] Wu Z X, Luo W Z, Chen C, et al. Research on Influencing Factors of Fuel Flow Based on QAR Data[M]. New York: Ieee, 2020: 800-804. [22] Ye L S, Cao L, Wang X H. EVALUATING FUEL CONSUMPTION FOR CONTINUOUS DESCENT APPROACH BASED ON QAR DATA[J]. Promet-Traffic & Transportation, 2019, 31(4): 407-421. [23] Wang H, Zuo H F, Sun J Z, et al. Research on On-line Monitoring Method of Lubricating Oil Consumption Rate of Aeroengine Based on QAR Data[M]. New York: Ieee, 2017: 191-197. [24] WU Ren-biao， CHEN Bin， SUN Shu-guang，et al. Fault diagnosis of airborne equipment simulationsystems based on slide window detection[J]. Journal of Civil Aviation University of China, 2012, 30(01): 1-5. [25] Li C Y, Sun J Z, Zuo H F, et al. Fault Detection for Air Conditioning System of Civil Aircraft Based on Multivariate State Estimation Technique[M]. New York: Ieee, 2017: 180- 185. [26] C. Edward LANa, WU Kaiyuanb, YU Jiang. Flight Characteristics Analysis Based on QAR Data of a Jet Transport During Landing at a High-altitude Airport [J]. Chinese Journal of Aeronautics, 2012, 25(01): 13-24. [27] Wang L X, Qian Y, Chen X G. Research on QAR Outliers Processing Based on Kalman Filtering and Newton Interpolation Algorithms[M]. New York: Ieee, 2020: 805-808. [28] SUN Rui-shan, YANG Yi-xuan. Research on Extraction of Key Parameters During Take-off Based on QAR Data [J]. China Transportation Review, 2015, 37(09): 58-63. [29] Uzun M, Demirezen M U, Koyuncu E, et al.: Deep Learning Techniques for Improving Estimations of Key Parameters for Efficient Flight Planning, 2019 Ieee/Aiaa 38th Digital Avionics Systems Conference, New York: Ieee, 2019. [30] Wang L, Ren Y, Wu C X. Effects of flare operation on landing safety: A study based on ANOVA of real flight data[J]. Safety Science, 2018, 102: 14-25. [31] Wang L, Zhang J Y, Dong C T, et al. A Method of Applying Flight Data to Evaluate Landing Operation Performance[J]. Ergonomics, 2019, 62(2): 171-180. [32] Liu S, Zhang Y X, Chen J T: A System for Evaluating Pilot Performance Based on Flight Data, Harris D, editor, Engineering Psychology and Cognitive Ergonomics, Cham: Springer International Publishing Ag, 2018: 605-614. [33] SUN Rui-shan，XIAO Ya-bing. Research on indicating structure for operation characteristic of civilaviation pilots based on QAR data[J]. Journal of Safety Science and Technology, 2012, 8(11): 49-54. [34] CAO Huiling, GAO Jianzhong, LIANG Damin. Research on Emission Diffusion Model of Civil Aircraft Engine in Cruise [J]. Environmental Science & Technology, 2014, 37(S1): 444- 447. [35] LI Chaoyi, SUN Jianzhong, YAN Hongsheng, et al. Estimation of exhausts pollution emissions for civil aircraft engine based on QAR data [J]. Chinese Journal of Environmental Engineering, 2017, 11(06): 3607-3616. [36] CAO Huiling, TANG Xinhao, MIAO Jiahe. Calculation and analysis of nitrogen oxide emission in LTO stage of engine 101 based on QAR data [J]. Acta,Science Circumstantiae, 2018, 38(10): 3900-3904. [37] CAO Huiling, LI Yuming, TANG Xinhao. Calculation and analysis of black carbon emissions from aircraft full flight phasebased on QAＲ data [J]. Acta Science Circumstantiae 2020, 40(06): 1951-1957. [38] Alizadeh A, Uzun M, Koyuncu E, et al. Optimal En-Route Trajectory Planning based on Wind Information[J]. Ifac Papersonline, 2018, 51(9): 180-185. [39] Lv H, Yu J J, Zhu T Y, et al. A Novel Method of Overrun Risk Measurement and Assessment using Large Scale QAR Data[M]. Los Alamitos: Ieee Computer Soc, 2018: 213-220. [40] Sun Rui-shan, Yang Yi-xuan, Wang Lei. Study on flight safety evaluation based on QAR data[J]. China Safety Science Journal, 2015, 25(07): 87-92. [41] Fang F, Zhang R Q, Zhao X B. An Aggregated Evaluation and Multi-dimensional Comparison Method of Flight Safety Based on QAR Data[M]. New York: Ieee, 2020: 145-149. [42] Wang X, Zhao X B, Yu L L. Data Mining on the Flight Quality of an Airline based on QAR Big Data[M]. New York: Ieee, 2020: 955-958.