Study On Fast Acoustic Source Localization Algorithm Based On Passive Detection

All kinds of applications based on target type and location continue to penetrate into all aspects of people’s life,so the design of high-precision real-time target location and recognition system has become a core hot topic for researchers at home and abroad.At present,it is an active research topic to identify the categories of sound events and high efficiency acoustic source localization.Aiming at the system of sound recognition and high efficiency acoustic source localization algorithm,combined with computer simulation and real data,this paper conducts in-depth research on passive detection and location algorithm based on signal processing.The contents are as follows:The present research results of sound recognition algorithm and passive sound source location algorithm are summarized and the advantages and disadvantages of these different algorithms are analyzed from various perspectives.In this paper,the existing classical methods to solve the location model of the time difference sound source are described,the key factors affecting the location accuracy of the model are analyzed.A deep learning based multi-overlapping sound event localization and detection algorithm in 3D space is proposed.Log-Mel spectrum and Generalized Cross-Correlation spectrum are joined together in channel dimension as feature input,and the features trained by neural network are input into classification and regression tasks in parallel to obtain recognition and localization results respectively.The channel attention mechanism is also introduced in the network to selectively enhance the features with large amount of information and suppress the useless features.Finally,a good result is obtained.Experiment shows that the proposed algorithm is robust to reverberation and environment,and achieves higher recognition and localization accuracy compared with the baseline method.Aiming at the low efficiency problem of the wide-band acoustic source localization algorithm based on the generalized cross-correlation inverse model,an acceleration algorithm is proposed.In the algorithm,in order to eliminate the microphone self-noise pollution,especially the self-noise pollution caused by the turbulent flow interacting with the microphone,we apply the diagonal remove(DR)method to the cross-spectral matrix(CSM),zeroing out its diagonal item.The computational grid is compressed by removing the grid points with low output power of microphone array.The geometric factor is also taken into consideration,the density-based clustering algorithm is used to further compress the grid,and finally only the reserved points are used for calculation to reduce the calculation scale.Finally,the propagation model matrix of the remaining point set is calculated and the Gauss-Seidel iterative method with positive constraints is used to solve the sound source position,so as to achieve the purpose of improving the calculation efficiency.Experiments with real data show that the proposed algorithm can effectively improve the computational efficiency.