Study On The Iterative Modified Orthogonal Matching Pursuit Deconvolution Beamforming Sound Source Identification Method

Accurate identification of sound source location is the precondition of noise control.Beamforming sound source identification method based on microphone array measurement because of its fast measurement speed,suitable for medium and long distance measurement,and high accuracy of steady state and transient sound source location,gradually become an indispensable source identification technology in the aviation,railway,automobile and other industries.However,conventional beamforming sound source identification methods have wide “main lobe”,poor spatial resolution and “sidelobe” contamination,which leads to the uncertainty of the identification results.Deconvolution beamforming sound source identification method can effectively improve the resolution and attenuate “sidelobe”,but because of its need to have multiple loop iteration solving process,which greatly reduce the computational efficiency,so most deconvolution beamforming algorithms have not yet been integrated into the commercial microphone array sound source identification system for engineering application.Orthogonal matching pursuit deconvolution approach for the mapping of acoustic sources(OMP-DAMAS),as a kind of sparse reconstructed deconvolution beamforming sound source identification method,has high spatial resolution and computational efficiency,and is suitable to be integrated into the microphone array sound source identification system.However,when OMP-DAMAS is used to identify multiple sound sources,the mutual interference between the sound sources leads to its inability to accurately identify the sound sources in some cases,and the need to construct a complete array point propagation function leads to a sharp decrease in the computational efficiency when the number of focus points is too much.Therefore,How to improve the existing omp-damas algorithm in the multi-source interference of the sound source identification performance to expand its application scenarios,improve the computational efficiency,and design a sound source identification system with both high resolution and high computational efficiency has become an urgent problem to be solved in the current sound source identification.In this thesis,the inverse problem of deconvolution beamforming sound source identification is firstly derived by formula.Then the basic theory and solving steps of OMP-DAMAS algorithm are given.After that the OMP-DAMAS algorithm and CB algorithm are compared through numerical simulation,and the performance of the two algorithms under different conditions is analyzed.Compared with CB algorithm,OMP-DAMAS algorithm can effectively improve the spatial resolution and attenuate the"side lobe"pollution,but the OMP-DAMAS algorithm is prone to errors in the identification of multiple sound sources when they interfere with each other,and the computational efficiency will decrease sharply as the number of focus points increases.In order to overcome the above defects of OMP-DAMAS algorithm,the thesis proposed an Iterative Modified OMP-DAMAS algorithm,which iteratively corrected the detected atoms and reduced the influence of the correlation interference between sound sources on the identification of multiple sound sources.Then,based on the shift invariance of the array point spread function(PSF),replaced the inner product of the iterative residual and PSF matrix in orthogonal matching pursuit solution process with cross-correlation function,and introduced fast Fourier transform to accelerate the solution process.The simulation results show that the proposed method can effectively solve the problem of multi-source interference,broaden the application scenarios,and have higher computational efficiency and practicability.The proposed algorithm is verified by the loudspeaker test,and the conclusion is basically consistent with the simulation result,which verifies the effectiveness and practicability of the proposed method.Finally,a sound source identification system is developed based on the general software development platform of visual studio C#.The system selected hardwares based on microphone array measurement method,presented the overall design and module design according to the actual needs of the sound source identification.The system designed seven modules which are the channel configuration module,the microphone calibration module,microphone array modeling module,data acquisition module,data analysis module,sound source identification module,data read and display module.With the functions of data acquisition,common acoustic analysis and sound source identification,the system provides a reliable tool for sound source identification.Some examples are given to verify the effectiveness of the system and the value of engineering application.