Research On Passive Acoustic Location Detection Method Based On Multi Element Acoustic Array

With the rapid development of array signal processing theory,sound source localization has become an active research field.The coexistence of far-field source and near-field source in real scene has been paid more and more attention.In this paper,multi microphone array is used to locate and detect indoor mixed sound source.Firstly,the research status of passive sound source location and detection methods at home and abroad is deeply studied,and the basic methods and principles of sound location are analyzed.By analyzing and comparing the performance of two classical microphone array localization algorithms,based on the generalized cross-correlation function,a generalized cross-correlation(GCC)method based on eigenstructure is proposed to estimate the time delay between microphones,and the main sound source vector is calculated from the frequency domain correlation matrix instead of the previous microphone The received signal is used for time delay estimation,and the properties of the linear equation model of sound source are analyzed,and the far-field model is derived in detail.Aiming at the problem of array aperture loss,a new algorithm for parameter estimation of far-field(FF)and near-field(NF)mixed narrow-band signal source based on sparse nested linear array and high-order cumulant is proposed.The fourth-order cumulant matrix is decomposed by singular value decomposition(SVD),and the direction of arrival(DOA)parameters are estimated by a rotation invariant(ESPRIT)algorithm.At the same time,in order to identify the NF and FF sources,a method to calculate the distance parameter of the mixed source is proposed,which directly separates the NF and FF sources.Aiming at the problem of low positioning accuracy of mixed sound source,a non iterative algorithm to complete cross spectral matrix is proposed.By using the Hermitian property of cross spectral matrix,without iteration,the computational efficiency is improved without sacrificing the positioning accuracy.Firstly,a sparse model is established,and a redundant impulse response(RIR)matrix is constructed as the measurement matrix of compressed sensing to transform the source localization problem into a compressed sensing problem.Then,according to the spatial sparse correlation of the direction vectors of multiple sources,the projection operator is introduced to keep the root mean square error of azimuth within 5% in the compressed sensing framework.The simulation results of uniform circular array(UCA)show that,compared with many traditional positioning methods,this algorithm has better positioning accuracy.