Research On Dynamic Modeling And Parameter Identification Methods Of Accelerometer Under Shock Excitation

The dynamic characteristics of accelerometer directly influence the dynamic performances of accelerometer based equipement and system.In the study of accelerometer’s dynamic characteristics,due to lacking dynamic model which can comprehensively describe the dynamic characteristics of accelerometer and corresponding parameters identificaiton methods,the improvement of measurement accuracy and dynamic performance of accelerometer based devices,instruements and control systems are limited.Mechanical shock has the characteristics of narrow pulse duration and wide frequency range and so on.It could stimulate the high-frequency dynamic response of accelerometer and is a typical excitation signal for studying the dynamic characteristics of accelerometers within a wide frequency range.Therefore,it is has important significance and practical value both in theoretical analysis and application domain that researching on the dynamic modeling and realizing dynamic model parameter identification of accelerometer under shock excitation.Based on the analysis of existing dynamic modeling and parameter identification methods of accelerometer,the dynamic modeling and parameter identification methods of accelerometer under shock excitation are researched.The influence factors of accelerometer dynamic modeling were researched.The estimated error of parameter that used current accelerometer dynamic model was given.And a hybrid structure dynamic model of accelerometer was proposed,which can describe the dynamic characteristics of accelerometer more comprehensively and provides a new way for researching the dynamic modeling of accelerometer.An accelerometer output signal de-noising method combining the improved empirical mode decomposition using second generation wavelets interpolation(SGWI-EMD)and average intrinsic mode function(aIMF)was proposed.This method obtained the aIMF by utilizing SGWI-EMD and applied spectral kurtosis(SK)as criterion,and then the noise could be seperated through an iterative way from the aIMF.Compared with the traditional signal de-noising methods based on linear filter or wavelets,the proposed de-noising method could remove the noise from accelerometer output signal effectively under shock excitation,proposing an effective method to reduce the influence of noise on the accuracy of dynamic model parameter identification of accelerometer.In order to reduce estimated error of accelerometer frequency response function(FRF),a FRF estimated method of accelerometer using discret spectrum correction and Hv(DSC-Hv)was given.This method could improve the frequency resolution with the length of original data unchanged,and reduce the influence of the fence effect and measurement noise effectively on the FRF;For the delay mismatch of mearsurement channels,a delay time estimated method using delay time correction and least square(DTC-LS)for accelerometer’s dynamic measurement was given.This method could obtain the delay time due to the delay mismatch in accelerometer dynamic measurement with high accuracy;Considering the influence of nonlinear further,a identification method for accelerometer dynamic model parameters using weight least square and support vector machine(WLS-SVM)was proposed.This method could reduce the influence of nonlinear on the accuracy of dynamic model parameter identification of accelerometer effectively with simple algorithm and good numerical stability.And it also improves the performance of anti-noise and enhances the robustness for unmodeling characteristics,which has important significance to promote the research on dynamic characteristics of accelerometerBased on virtual instrument technology,an integration method of the dynamic model parameter identification system for accelerometer was given.This system could generate shock excitation with a wide frequency range more than 50 kHz;and the software architecture of the system was analyzed.Based on the integrated frame of file based modularization and hybrid programming technology,the software of the system was completed.The software could be stand-alone installed and run program separated from MATLAB and Lab VIEW environment with strong generality,making a positive exploration for the development of sensor dynamic testing instrument and system with high value in engineering application domain.In order to verify the efficiency of the proposed methods,the established accelerometer dynamic model parameter identification system was applied to the calibration experiements of accelerometer upon shock excitation.The results of experiments show the proposed dynamic modeling and parameters identification methods of accelerometer are feasible and the dynamic model parameters of accelerometer could be obtained with higher accuracy.It could significantly promote the process of engineering application for the dynamic modeling and parameter identification methods of accelerometer.The proposed hybrid structure dynamic modeling method could comprehensively describe the accelerometer’s dynamic characteristics,and provides a new way for researching the dynamic modeling of accelerometer.The proposed dynamic model parameter identification method of accelerometer could estimate the dynamic model parameters of accelerometer with high accuracy by reducing the influences of noise,delay mismatch of mearsurement channels and nonlinear,and has wide application prospect in the accelerometer dynamic calibration field.