Research Result For Direction Of Arrival Estimation Based On MEMS Acoustic Vector Sensors

The MEMS vector hydrophone has the advantages of small size,well consistent,low cost and so on,which can obtain a scalar acoustic pressure information and three orthogonal particle acceleration vector acoustic information about a location in the sound field,the multi-channel output signal also provides more flexibility for the detection of the sound source signal and the direction of arrival(DOA)estimation,and has the ability to cancel the isotropic noise interference.The vector hydrophone technology has been widely concerned by the marine power in the field of underwater target detection,submarine stealth,sound source orientation and other fields,and has stable performance of vector hydrophone into the engineering application stage.The problem based on the vector hydrophone has become a frontier hot topic in the research of all countries.In this paper,we systematically studied the adaptive algorithm and DOA estimation algorithm in the underwater acoustic signal processing.The main research contents are as follows:The first part: According to the difference of the characteristics of the underwater acoustic signal and the noise signal,an improved algorithm based on the combination of LMS adaptive noise cancellation and Fourier transform filtering is proposed for the problem of ?submerged? data collected by MEMS hydrophone in strong noise field.Then from the mean square error(RMSE),the signal-to-noise ratio(SNR),SNR gain(GSNR)and Resembler(R)to evaluate the denoising performance between the filter signal and the ideal signal.The simulation results show that the improved algorithm breaks the limitation of the traditional adaptive noise cancellation method in the low signal-to-noise ratio(less than 0dB)and the non-stationary noise environment.When the SNR falls below-15 dB,the improved algorithm still has a good ability to deny.The second part: The improved denoising algorithm is used to obtain the pretreatment signal,four kinds of algorithms which contain the arc-tangent method,average acoustic intensity method,cross spectral algorithm and least square method are used to estimate the direction of arrival(DOA,Direction Of Arrival)of a single vector hydrophone.The estimation results are compared with the preset azimuths,and the mean square error and success rate of the improved algorithm and the center filtering algorithm are analyzed,evaluation of two kinds of algorithms in different snapshots and SNR performance environment,to provide experimental basis for engineering application.The third part: The improved DOA estimation of the vector hydrophone array is carried out by using the improved algorithm combined with the multiple signal classification(MUSIC)algorithm.The advantages and disadvantages of the improved MUSIC algorithm and the original MUSIC algorithm are compared and analyzed respectively in different snapshots and SNR performance environment,including the mean square error and the prediction success rate of the azimuth estimation.It is shown that the improved algorithm in the condition of low SNR still has certain engineering application.The simulation results show that the improved denoising algorithm greatly improves the accuracy of DOA estimation.The average sound intensity method and the cross spectrum method also have a good accuracy of DOA estimation in a small number of snapshots and lower SNR performance environment based on a single-vector hydrophone;the improved MUSIC algorithm has higher prediction success rate and smaller mean square error value of azimuth than the original MUSIC algorithm under the small snapshots and low SNR.The fourth part: Finally,the Lake experiment at second Fenhe Reservoir was carried out.Based on the experimental data of Fenhe lake,it is shown that the results of data preprocessing show good results,which include good performance,simple and efficient calculation,and provide satisfactory effects for practical application.