Signal Denoising Method Of MEMS Vector Hydrophone Based On Wavelet Threshold Processing

Underwater and acoustic environments such as oceans and lakes are complex and changeable.The signals received by MEMS vector hydrophones often contain complex environmental noises,which makes the received signals inevitably produce signal distortion and baseline drift.The angle and position of the sound source signal deviate greatly from the actual value.In view of this kind of phenomenon,this paper studies the signal denoising method of MEMS vector hydrophone based on wavelet threshold processing.The main research contents are as follows:(1)A soft threshold wavelet denoising method based on improved genetic algorithm is proposed.Starting with the method of obtaining the optimal wavelet threshold,the genetic algorithm can be used to quickly search for the optimal wavelet threshold,and then use the soft threshold wavelet method to denoise the signal.In this method,considering the problems of the traditional genetic algorithm and the existing improved genetic algorithm,the genetic operator and the sequence of genetic operations are adaptively changed according to the degree of population fitness dispersion to obtain the optimal wavelet threshold.Both simulation experiments and fenji measured data experiments conducted in 2011 in the Fenhe Second Reservoir of Taiyuan in North University of China show that this method can effectively remove the noise contained in the signal and has certain superiority and practicality.(2)A signal denoising and drifting method based on VMD and nonlinear wavelet threshold processing is proposed.In this part,the wavelet threshold function is improved,so that it can be adaptively transformed according to the number of layers of wavelet decomposition,and then achieve a better noise filtering effect.After the variational modal decomposition is performed on the signal,the IMFs obtained by the decomposition is divided into three parts: the noise IMF component,the noisy IMF component,and the pure IMF component.The basis is mainly the magnitude of the correlation coefficient of each component.Finally,the noisy IMF component is denoised by an improved threshold function after wavelet transform,and finally reconstructed with the pure IMF component to obtain the overall denoised signal.The simulation experiments and the fenji measured data experiments carried out in the Fenhe Second Reservoir of Taiyuan in 2011 and 2014 by North University of China both show that the method can effectively remove the noise and baseline drift phenomena contained in the signal,which has certain advantages and Practicality.This paper proposes two new signal processing methods,both of which can effectively remove noise.Especially the signal denoising and drifting method based on VMD and nonlinear wavelet threshold processing can also effectively remove the baseline drift phenomenon of the signal,which can lay a certain foundation for the further azimuth estimates.