Jtam-A4.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 54, 3, pp. 1025-1037, Warsaw 2016 DOI: 10.15632/jtam-pl.54.3.1025 MODELLING OF THE VIBRATION EXPOSURE IN TYPICAL WORKING MACHINES BY MEANS OF RANDOM INPUT SIGNALS Igor Maciejewski, Tomasz Krzyżyński Koszalin University of Technology, Faculty of Technology and Education, Koszalin, Poland e-mail: igor.maciejewski@tu.koszalin.pl; tomasz.krzyzynski@tu.koszalin.pl The aim of the paper is to formulate a generalizedmethodology ofmodelling randomvibra- tions that are experienced by machine operators during their work. In the following paper, spectral characteristics of input vibrations are specified in such a way that the generated excitation signals are representative for different types of working machines. These signals could be used for determination and evaluation of risks from exposure to whole-body vibra- tion. Keywords: simulated input vibration, random vibration 1. Introduction There are two basic sources of mechanical vibrations that can disturb proper functioning of machines. The first one relates to systems that during their operation generate vibrations, e.g. engine vibrations in the workingmachine (Preumont, 2002). The second class concerns systems thatvibratedue to external factors, e.g. cabvibrations of themachine that ismoving over uneven ground (Maciejewski andKrzyżyński, 2011;Nabaglo et al., 2013).Many sources of vibrations can cause periodic or random reactions ofmachine elements (Sapiński andRosół, 2007;Maślanka et al., 2007). In such a situation, resonant states can be obtained and theymay cause disturbances inmotion of individualmachine elements.Resonant vibrations have an adverse effect onmachine functioning, and this can lead to their failure. In addition, vibrations have a negative influence on health of operators of workingmachines. Machine operators during their work are exposed to vibrations that are caused very often by motion of machinery over uneven ground (Kowal et al., 2008; Snamina et al., 2013). In the case of whole-body vibration, the seat of a vehicle or the platform of a worker vibrates and this motion is transmitted into the humanbody.Under normal operating conditions, humans occupy the following body positions (Griffin et al., 2006): • sitting position (mechanical vibrations transmitted through pelvis) that precludes active damping of vibration using lower limbs, • standingposition (mechanical vibrations transmitted through feet) that allows active dam- ping of vibration in the low frequency range. The basic positions of the human body at work are presented in Fig. 1. Harmful vibrations can be transmitted to the human body in three orthogonal directions: longitudinal x, lateral y and vertical z. Vibrations are defined by their magnitudes and frequencies. The magnitudes of vibration are traditionally expressed as vibration acceleration, because most vibration transdu- cers produce an output that is related to the acceleration signal. Frequencies appearing in the random whole-body vibration environment usually occur between 1 and 20Hz (Krzyzynski et al., 2004). Exposure to whole-body vibration causesmotions and forces within the human body that may cause discomfort, adversely affect working performance and cause health and safety risk (Griffin et al., 2006). 1026 I. Maciejewski, T. Krzyżyński Fig. 1. Positions of the human body at work: sitting position (a), standing position (b) Thebasic opportunity tominimise harmful vibrations consists in applying a vibration reduc- tion system that prevents propagation ofmechanical vibrations from their source to the isolated body (Kowal et al., 2008). Dynamic characteristics of a vibration isolator should be selected for a specific application in such a way that the vibrations transmitted from the source to the body are minimal (Snamina et al., 2013). In order to analyze dynamics of a vibration isolation system, the excitation signal with specific spectral characteristics should be used. Unfortuna- tely, there are no effective methods for generating signals representing the working of different machinery. In the papers [3], [7], [8], the target spectral characteristics are only standardised for the simulated input vibration test in the vertical direction. There is a lack of effective procedu- res of generating the time history of random signals with precisely defined spectral properties. In addition, it is not known how to reproduce the excitation signals based on the measured spectral characteristics (Bluthner et al., 2008), especially in horizontal directions (lateral and longitudinal). In the following paper, an effective method of reproducing random vibration in different types of working machines is proposed for the purpose of selecting dynamic characteristics of vibration reduction systems.At first,machines having similar operational vibration are grouped by virtue of various mechanical characteristics. Then a signal generator is utilized to generate random vibration and an original signal processing technique is used in order to obtain the spectral characteristics representing vibration in a specific group of machines. 2. Modelling of the vibration reduction system The general model of the vibration reduction system used in typical working machines is pre- sented in Fig. 2. The suspended body is isolated against harmful vibrations in three orthogonal directions: longitudinal x, lateral y and vertical z. The passive visco-elastic elements are used in order tominimise thehumanexposure towhole-bodyvibration.Rotation vibrations aroundeach axis of the Cartesian coordinate system (x,y,z) are neglected in this simplifiedmodel. Ongoing research (Maciejewski et al., 2011) indicates that the exposure of workers to risks arising from vibration are evaluated for translational axes, therefore in this paper, vibro-isolation properties are discussed for this specific direction. The equation of motion for such a system is formulated in the matrix form Miq̈i+Diq̇i+Ciqi =Fsi i=x,y,z (2.1) Modelling of the vibration exposure in typical working machines... 1027 where qi is the displacement vector of the isolated body, Mi, Di, Ci are the inertia, damping and stiffness matrices, respectively, Fsi is the vector of exciting forces describing the non-linear vibration isolator. Fig. 2. General model of the non-linear vibration reduction system used in typical workingmachines There are dozens of human body models presented in the modern literature (Rutzel et al., 2006; Stein et al., 2007; Toward and Griffin, 2011). Usually, there are multi-degree of freedom lumped parameter models that consider the seating and standing position. In this paper, the generalised mathematical model of vibration reduction system is presented that allows one to use various bio-mechanical structures of the well-known human body models. The n-element vector qi represents the movement of elements contained in the bio-mechanical model of the human body qi = [q1i,q2i, . . . ,qni] T i=x,y,z (2.2) where: q1i,q2i, . . . ,qni are displacements of the human body model. The n-element vector of exciting forces Fsi is given by the following expression Fsi = [Fsi,0, . . . ,0] T i=x,y,z (2.3) The particular non-linear exciting forces Fsi can be described in a general form as follows Fsi = l∑ k=1 Fdik(q̇1i− q̇si)+ l∑ k=1 Fcik(q1i− qsi) i=x,y,z (2.4) where: Fcik(q1i−qsi) are non-linear functions including force characteristics of the conservative elements as a function of the system relative displacement q1i − qsi, Fdik(q̇1i − q̇si) are non- -linear functions including force characteristics of the dissipative elements as a function of the system relative velocity q̇1i− q̇si. These non-linear characteristics should be selected especially for well-defined input vibrations by using appropriate element models. Exemplary models can be found in the authors’ previous paper (Maciejewski et al., 2011). The input displacement qsi and velocity q̇si aremodelled as discrete-time random signals that are generated for the specific direction of vibration exposure (i=x,y,z). Modelling of the random vibration acceleration q̈si(t) occurring in typical workingmachines is presented in the next part of the following paper. The elaborated signal models allow one to 1028 I. Maciejewski, T. Krzyżyński test a variety of vibration reduction systems by means of the laboratory measurements and/or numerical simulations. The proposed method can be used for generating signals specified by the present test standards, see [3], [7], [8] and also for reproducing signals measured by other authors (Bluthner et al., 2008).The requirements regarding limitations of thevibration simulator (shaker) can be found in International Standards ([3], [7], [8]). Safety requirements for the test person are exhaustively discussed in ISO-2613 [6]. 3. Random signal generator The simulated input vibration is generated using a normally (Gaussian) distributed random numbers (rand) that produces a waveform q̈sij(t)= √ σ2sijrand (tk ts ) i=x,y,z j=1, . . . , l (3.1) where:qsij(t) is vibrationacceleration generated in thedirectionx,y,z, j=1, . . . , l is thenumber of signal generator, σ2sij is the variance of random numbers, t is the current time instant, tk is the computation time, ts is the time interval between samples. In order to simulate the effect of white noise with a correlation time near 0 and a flat power spectral density, a sampling time should be much smaller than the fastest dynamics of the system. For intended analysis with high resolution, an acceptable random signal can be achieved by specifying the sampling frequency 100 times (or more) of the maximal frequency of the system to be tested. Therefore, the time interval between samples of the random signal is ts =0.01/fmax, where fmax is the maximal frequency of the system in Hz. For generation of vibration up to 20 Hz, the maximal frequency of the random signal is set to 25Hz. The probability density function of a normally distributed random signal can be described using the following relation p(q̈sij(t))= 1√ 2πσ2sij exp ( − 1 2σ2sij q̈2sij(t) ) i=x,y,z j=1, . . . , l (3.2) The time plot of a stochastic signal q̈sij(t) is shown in Fig. 3a, and its histogram on the back- ground of the probability density function is illustrated in Fig. 3b. Fig. 3. Input vibration generated as a normally distributed random process (a) and its histogram (grey box) on the background of the probability density function (black line) (b) The spectral properties of the generated random signal have to be close to white noise, so the power spectral density (PSD) should be flat in the considered frequency range, and the correlation time of a time series shall be close to zero (Bendat and Piersol, 2004). The power spectral density and the normalized auto-correlation of the generated signal are presented in Fig. 4. Modelling of the vibration exposure in typical working machines... 1029 Fig. 4. Power spectral density (a) and normalized autocorrelation (b) of the generated random signal 4. Signal processing technique An original signal processing technique is proposed in order to obtain specific spectral proper- ties of the generated random signal. This technique involves the making use of a set of the Butterworth filters (high-pass and low-pass). A block diagram of the proposed signal processing technique is shown in Fig. 5. Fig. 5. Block diagram of the proposed signal processing technique Transfer functions for the linear Butterworth filters are defined in the following way (Parks and Burrus, 1987): — high-pass filter GHPsij(s)= sn sn+an−1sn−1+ . . .+a1s+1 i=x,y,z j=1, . . . , l (4.1) — low-pass filter GLPsij(s)= 1 sn+an−1sn−1+ . . .+a1s+1 i=x,y,z j=1, . . . , l (4.2) where: a1 to an are the coefficients of the Butterworth filter specified in Table 1, n is the order of the filter, s is the Laplace variable, j=1, . . . , l is the number of filters used. 1030 I. Maciejewski, T. Krzyżyński Table 1.Coefficients of the Butterworth filter a1, . . . ,an for different filter orders n n a1 a2 a3 a4 a5 a6 a7 2 1.414 – – – – – – 3 2.000 2.000 – – – – – 4 2.613 3.414 2.613 – – – – 5 3.236 5.236 5.236 3.236 – – – 6 3.863 7.464 9.141 7.464 3.863 – – 7 4.493 10.097 14.591 14.591 10.097 4.493 – 8 5.125 13.137 21.846 25.688 21.846 13.137 5.125 The high-pass and low-pass filters (Eqs. (4.1) and (4.2)) allows one to create the required band-pass filter for a specified frequency bandwidth. A combination of the high-pass and low- -pass filters at the defined cut-off frequency and filter order provides generating of the input vibrationwith various spectral characteristics. The transfer function of the total filter is defined as follows Gsi(s)= Nsi(s) Dsi(s) = l∑ j=1 GHPsij(s)GLPsij(s) i=x,y,z j=1, . . . , l (4.3) where: Nsi(s) and Dsi(s) are the numerator and denominator polynomials that represent a combination of the Butterworth filters shown in Fig. 5. The transfer function (Eq. (4.3)) is especially useful when analyzing the proposed filter stability. If all poles of this transfer function have negative real parts, then the filter is considered as asymptotically stable. However, the filter is unstable when any pole has a positive real part. Therefore, all poles in the complex s-plane should be located in the left half plane to ensure filter stability, which can be described by the following relation Re(Dsi(s))< 0 i=x,y,z (4.4) Pole distribution in the complex s-plane for unstable and stable filter is shown in Fig. 6. Fig. 6. Pole distribution in the complex s-plane: unstable filter (a), stable filter with stabilitymargin (b) The input vibration is defined by a power spectral density of the vertical acceleration and by the root mean square value of the acceleration signal. The target magnitude PSDsi(2πf) of the power spectral density function can be calculated with the assistance of the following equation PSDsi(2πf)= l∑ j=1 2σ2sij fs ∣∣∣(GHPsij(2πf))2(GLPsij(2πf))2 ∣∣∣ i=x,y,z j=1, . . . , l (4.5) Modelling of the vibration exposure in typical working machines... 1031 where: f is the frequency in Hz, σ2sij is the signal variance in a specified frequency bandwidth, fs is the sampling frequency. The curve defined by Eq. (4.5)) consists of target power spectral density to be produced for the simulated input vibration (Fig. 7). Fig. 7. Target power spectral density of the simulated input vibration The power spectral density of the simulated acceleration signal is considered to be represen- tative for different types of working machines if and only if ([3], [7], [8]): • the magnitude of simulated input is within the tolerance of the target power spectral density function PSDsi(2πf)±0.1maxf(PSDsi(2πf)), • the root mean square (RMS) value of simulated input acceleration is within the tolerance of the required value (q̈si)RMS ±0.05(q̈si)RMS. The root mean square value of the acceleration signal is defined using the following relation (q̈si)RMS = √√√√√ 1 tk tk∫ 0 (q̈si(t))2 dt i=x,y,z (4.6) where: q̈si(t) is the time history of input vibration and tk is the durationwithin which vibration data for analysis is obtained. 5. Spectral estimation method A novel method is proposed in order to evaluate parameters (signal variances and filter cut-off frequencies) of the system presented in Fig. 5. Such a parametric method allows one to find the system configuration with a magnitude response approximating the desired function (Fig. 8). Therefore, a random signal with the user-defined spectral properties can easily be generated. This is the essence of the importance and novelty of the authors’ method in comparison with the spectral estimation methods used by most authors. The proposed method applies the least square error (LSE)minimization technique over the frequency range of the filter response. 1032 I. Maciejewski, T. Krzyżyński Fig. 8. Magnitude response (—) approximating the desired function (- - -) The least square error of the specified frequency response is described in the following form LSEsi = √√√√ m∑ k=1 ( PSDsi(2πfk)− P̂SDsi(2πfk) )2 i=x,y,z (5.1) where:PSDsi and P̂SDsi are thedesired and estimatedpower spectral densities that are obtained for the same frequency value fk. The error minimization includes design parameters having an influence on the spectral cha- racteristics of the generated signal as follows min σ2 sij ,fHPsij,fLPsij LSEsi(σ 2 sij,fHPsij,fLPsij) i=x,y,z j=1, . . . , l (5.2) where: σ2sij is the variance of the random signal, fHPsij and fLPsij are the cut-off frequencies of the high-pass and low-pass Butterworth filters, respectively. The problem ofminimizing the objective function (Eq. (5.2)) of several parameters (decision variables) is defined subject to linear inequality constraints on these variables (σ2sij)min ¬σ 2 sij ¬ (σ 2 sij)max (fHPsij)min ¬ fHPsij ¬ (fHPsij)max i=x,y,z j=1, . . . , l (fLPsij)min ¬ fLPsij ¬ (fLPsij)max (5.3) where: (σ2sij)min and (σ 2 sij)max are the lowest andhighest valueof the signal variance, (fHPsij)min and (fHPsij)max are the lowest andhighest value of the cut-off frequencies of the high-pass filter, (fLPsij)min and (fLPsij)max are the lowest and highest value of the cut-off frequencies of the low-pass filter. Additionally, nonlinear inequality constraints are imposed on the objective function (Eq. (5.2)) to restrict the filter cut-off frequencies in the following order fHPsi1 ¬ fLPsi1 ¬ fHPsi2 ¬ fLPsi2, . . . ,¬ fHPsij ¬ fLPsij i=x,y,z j=1, . . . , l (5.4) where: j = 1, . . . , l is the number of low-pass (LP) and high-pass (HP) filters used to estimate the input vibration in different directions (i=x,y,z). Modelling of the vibration exposure in typical working machines... 1033 Themultiple correlation coefficient (R) is employed in order tomeasure how successful is the approximation in explaining the variation of the optimisation results. This coefficient is defined as follows Rsi = √√√√√√√1− ∑m k=1 ( P̂SDsi(2πfk)−PSDsi(2πfk) )2 ∑m k=1 ( PSDsi(2πfk)−PSDsi(2πfk) )2 i=x,y,z (5.5) where: P̂SDsi is the estimated power spectral density for the directions x, y, z, PSDsi is the desired power spectral density, PSDsi is the mean value of the desired power spectral density, fk is the discrete value of frequency. The coefficient Rsi can take any value between 0 and 1, with a value closer to 1 indicating a better estimation of the power spectral density. 6. Vertical vibration The standards ([3], [7], [8]) specify the laboratory simulated vertical vibration (z-axis) that is based on representative measured data from different types of machines in typical working con- ditions.The input spectral classes are defined formachines having similarmechanical characteri- stics. The test inputs indicated in these standards are based on a large number ofmeasurements taken in situ of working machines while they were used under severe operating conditions. International Standard ISO7096 [7] specifies input vibrations for earth-movingmachinery in nine spectral classes (EM1-EM9). British Standard BS EN 13490 [3] defines the input spectral classes required for industrial trucks (IT1-IT4). ISO 5007 standard [8] specifies input vibration in three input spectral classes (AG1-AG3) for agricultural tractors with rubber tyres, unsprung rear axles and no low-frequency cab isolation. Each class defines a group of machines having similar vibration characteristics. All of these standards completely designate high-pass and low-pass filters of theButterworth type that are required for generating vibration along the vertical axis. Numerical values of the cut-off frequencies and filter orders are presented in Table 2. The vibration characteristics for selected input spectral classes, i.e. EM1, EM5-EM7, IT2-IT3, AG1-AG2 and their tolerances, are shown in Fig. 9. The desired and obtained rootmean square values of the acceleration signal are defined in Table 2. Table 2. Parameters of the input vibration generated for different types of working machines Signal generators High-pass filters Low-pass filters Results σ2sij, [( m s 2)2] fHPsij [Hz] nHPsij fLPsij [Hz] nLPsij (q̈si)RMS, [ m s 2] Type of Input i j j j j j Desi- Obta- machines vibration 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 red ined Earth EM1 z 282 – – 1.5 – – 4 – – 2.5 – – 4 – – 1.71 1.67 moving EM5 z 111 – – 1.5 – – 4 – – 3.5 – – 1 – – 1.94 2.10 machinery EM6 z 79 – – 6.5 – – 2 – – 9 – – 2 – – 1.65 1.70 EM7 z 925 – – 3 – – 8 – – 3.5 – – 8 – – 2.25 2.27 Industrial IT2 z 145 – – 3 – – 4 – – 3 – – 2 – – 1.05 1.05 trucks IT3 z 60 – – 1.5 – – 4 – – 3 – – 4 – – 0.96 0.94 Agricultural AG1 z 925 – – 3 – – 8 – – 3.5 – – 8 – – 2.26 2.27 tractors AG2 z 722 – – 2.1 – – 8 – – 2.6 – – 8 – – 1.94 1.91 Agricultural AT1 x 335 51 27.1 2 5.2 9.7 8 4 4 2.4 5.7 12 4 4 4 1.62 1.69 tractor y 728 165 27.3 1.1 3 7.7 8 4 4 1.3 3.4 10 8 4 4 1.86 1.84 Articulated AL1 x 7.9 6,8 3.7 0.2 1.3 5.7 8 4 4 1.2 2.3 7 8 4 2 0.50 0.49 truck y 15.7 21.9 1.8 0.4 2 6 4 8 4 1.5 2.4 14 4 8 2 0.62 0.63 1034 I. Maciejewski, T. Krzyżyński Fig. 9. Power spectral densities of simulated vertical input vibration (—) for spectral classes: EM1 (a), EM5 (b), EM6 (c), EM7 (d), IT2 (e), IT3 (f), AG1 (g), AG2 (h) and their tolerances (- - -) 7. Horizontal vibration In contrary to vertical vibration, there are no standardised signals available for the horizontal directions (x-axis andy-axis).Therefore, thispaper specifies inputvibrations for selected spectral classes and each class consisting of longitudinal and lateral cabin floor vibration for particular working machines performing a specific operation. In the paper by Bluthner et al. (2008), the test inputs were measured as a part of the study that has been done within the framework of the European research project VIBSEAT. They do not have sufficient magnitudes to cover the majority of actual spectra observable on the field. The determination of representative spectra represents a large amount of work which is not the purpose of this paper. Modelling of the vibration exposure in typical working machines... 1035 There are no designated Butterworth filters that are desired for generating vibration along the horizontal axes. Therefore, at first, PSD functions of these signals are determined using the spectral estimation method shown in Section 5. Then the variance of random signals and the cut-off frequencies of filters are evaluated using the objective function described by Eq. (5.2). Numerical values of the design parameters are presented in Table 2. The spectral characteristics of measured, estimated and simulated input vibrations, i.e. AT1, AL1 and their tolerances for the x and y axes, are shown in Fig. 10. The desired and obtained root mean square values of the acceleration signal are defined in Table 2. Fig. 10. Power spectral densities of measured (- - - ), estimated (—) (left-hand side) and simulated (—) (right-hand side) horizontal input vibrations for spectral classes: AT1 x-axis (a), (b), AT1 y-axis (c), (d), AL1 x-axis (e), (f), AL1 y-axis (g), (h) and their tolerances (-·-·-) 1036 I. Maciejewski, T. Krzyżyński 8. Conclusions In this paper, the theoretical models of simulated input vibrations are developed in such a way that the specific spectral properties of signals are obtained using the original filtration technique. This is thebasis of theproposedprocedure for generating randomvibration that is representative for vibrations affecting the operators at work. The models developed in this paper can be used to reproduce the real working conditions of vibration reduction systems in different types of machines. Investigations carried out using the proposed signals should assist a selection process of vibro-isolation properties of systems for different spectral classes of the input vibrations. Acknowledgements The following work has been a part of the research project “Methods and procedures of selecting vibro-isolationpropertiesofvibration reductionsystems” fundedby theNationalScienceCenter ofPoland under the contract No. UMO-2013/11/B/ST8/03881. References 1. Bendat J.S., Piersol A.G., 2004, Methods for Analysis and Measurement of Random Signals (in Polish), Polish Scientific Publishers PWN,Warsaw 2. Bluthner R., Seidel H., Hinz B., 2008, Laboratory study as basis of the development for a seat testing procedure in horizontal directions, International Journal of Industrial Ergonomics, 38, 447-456 3. British Standards Institution BS EN 13490, 2002, Mechanical vibration – Industrial trucks – La- boratory evaluation and specification of operator seat vibration, London 4. Directive 2002/44/EC of the European Parliament and of the Council, 2002, On the minimum health and safety requirements regarding the exposure of workers to the risks arising fromphysical agents (vibration), Official Journal of the European Communities, 13-18 5. Griffin M.J., Howarth H.V.C., Pitts P.M., Fischer S., Kaulbars U., Donati P.M., BeretonP.F., 2006,Guide to goodpractice onwhole-body vibration.Non-binding guide to good practice with a view to implementation of Directive 2002/44/EC on the minimum health and safety requirements regarding the exposure of workers to the risks arising from physical agents (vibrations), European Commission, Luxembourg 6. International Organization for Standardization, 1997,Mechanical vibration and shock – Evolution of human exposure to whole body vibration, ISO 2631, Genewa 7. International Organization for Standardization, 2000, Earth-movingmachinery – Laboratory eva- luation of operator seat vibration, ISO 7096, Genewa 8. International Organization for Standardization, 2003, Agricultural wheeled tractors – Operator’s seat – Laboratorymeasurement of transmitted vibration, ISO 5007, Genewa 9. Kowal J., Pluta J., Konieczny J., Kot A., 2008, Energy recovering in active vibration isola- tion system – results of experimental research, Journal of Vibration and Control, 14, 7, 1075-1088 10. Krzyzynski T., Maciejewski I., Chamera S., 2004, Modeling and simulation of active sys- tem of truck vibroisolation with biomechanical model of human body under real excitation, VDI Publications (The Association of German Engineers), 1821, 377-390 11. Maciejewski I., Kiczkowiak T., Krzyzynski T., 2011, Application of the Pareto-optimal ap- proach for selecting dynamic characteristics of seat suspension systems,Vehicle SystemDynamics, 49, 12, 1929-1950 12. Maciejewski I., Krzyżyński T., 2011,Control design of semi-active seat suspension, Journal of Theoretical and Applied Mechanics, 49, 4, 1151-1168 Modelling of the vibration exposure in typical working machines... 1037 13. Maślanka M., Sapiński B., Snamina J., 2007, Experimental study of vibration control of a cable with an attachedMRdamper, Journal of Theoritical and Applied Mechanics, 45, 4, 893-917 14. Nabaglo T., Kowal J., Jurkiewicz A., 2013, Construction of a parametrized tracked vehicle model and its simulation in MSC.ADAMS program, Journal of Low Frequency Noise Vibration and Active Control, 32, 1/2, 167-173 15. Parks T.W., Burrus C.S., 1987,Digital Filter Design, JohnWiley & Sons, NewYork 16. Preumont A., 2002, Vibration Control of Active Structures An Introduction, Kluwer Academic Publishers, London 17. Rutzel S., Hinz B., Wolfel H.B., 2006, Modal description – a better way of characterizing human vibration behavior, Journal of Sound and Vibration, 298, 810-823 18. Sapiński B., Rosół M., 2007, MR Damper performance for shock isolation, Journal of Theore- thical and Applied Mechanics, 45, 1, 133-145 19. Snamina J., Kowal J., Orkisz P., 2013,Active suspension based on lowdynamic stiffness,Acta Physica Polonica A, 123, 6, 1118-1122 20. Stein G.J, Muka P., Chmurny R., Hinz B., Bluthner R., 2007, Measurement and model- ling of x-direction apparent mass of the seated human body - cushioned seat system, Journal of Biomechanics, 40, 1493-1503 21. Toward M., Griffin J., 2011, The transmission of vertical vibration through seats: influence of the characteristics of the human body, Journal of Sound and Vibration, 330, 6526-6543 Manuscript received April 28, 2015; accepted for print February 16, 2016