Research On Denoising And Parameter Identification Of Accelerometer Signal Based On Wavelet Transform

Accelerometers are sensors used to measure the acceleration of the moving direction of the carrier.Because of their convenience and reliability,the accelerometers are widely used in aerospace,architecture,industrial automation and other fields.The dynamic model of accelerometer provides the basis for the measurement of acceleration signal.Realizing the parameter identification of the model can describe the dynamic characteristic of the accelerometer,and thus improves the precision of the acceleration measurement signal.However,the accelerometer signal is easily affected by the noise and low frequency nonlinearity,and the accuracy of the parameter identification of the accelerometer dynamic model is restricted by the noise based accelerometer signal.Therefore,the accuracy of filtering the noise in the accelerometer signal can effectively improve the precision of the signal analysis and the parameter identification of the dynamic model of the accelerometer.The existing signal denoising method based on empirical mode decomposition can decompose the signal adaptively,however,there are endpoint effect and modal aliasing in the process of noise reduction,which lead to low noise reduction.The signal de-noising method based on wavelet transform obtains the noise part of the measurement signal from the multi-resolution analysis measurement signal and realizes effective filtering.However,this method is overly dependent on the singleness of the measured signal,and does not fully consider the correlation between the accelerometers’output signals in the accelerometer’s comparison method,which restricts the improvement of the noise reduction precision.Therefore,it is of great theoretical significance and practical application value to study research on denoising and parameter identification of accelerometer signal based on wavelet transform.Based on the analysis of the relationship between the correlation coefficient of the wavelet coefficients of the accelerometer output signal and the noise,a noise reduction method for accelerometer signals based on sequential correlation and wavelet transform is proposed.By using the correlation between the wavelet coefficients of different accelerometer response signals under the same excitation,the cross correlation coefficient is introduced into the threshold calculation process of wavelet threshold denoising.This method can effectively reduce the noise in accelerometer output signal and improve the estimation accuracy of the frequency response function of accelerometer.A frequency domain identification method for accelerometer dynamic model parameters based on wavelet denoising and weighted least squares is proposed.This method uses the wavelet denoising signal to get the frequency response,and then uses the frequency response near the resonant frequency of the accelerometer to identify the parameters.In the same time,this method determines the complexity of sequences by arranging entropy,and adjusts the weight matrix of weighted least squares.This method can effectively reduce the influence of measurement noise and low frequency nonlinearity on parameter identification results,and improve the accuracy of parameter identification.The MATLAB platform simulation data and accelerometer comparison method of the impact excitation calibration show that the noise reduction method based on sequential correlation and wavelet transform can effectively suppress the noise of accelerometer signals and the proposed method based on wavelet denoising and weighted least square frequency domain parameter identification can reduce the influence of measurement noise and nonlinearity on the parameter identification process of dynamic model,and has a high precision of parameter identification.