Algorithm Improvements Of Sound Source Identification And Sound Field Separation With Equivalent Source Method Based On Sparse Regularization

Noise source location is a prerequisite for controlling vehicle noise and improving NVH performance.Near-field acoustic holography is an efficient technology to realize noise source visualization through microphone measurement signals,and is widely used in automotive and fault diagnosis fields.The near-field acoustic holography based on the equivalent source method has become a hot research topic because it is easy to implement and suitable for identifying sound sources with complex surface shapes.The core difficulty is the solution of equivalent source intensity.In addition,there are often interference sources or reflective surfaces near the target sound source in engineering practice.In order to realize the accurate identification of the target sound source,the sound field of the target sound source and the interference source need to be separated by sound field separation technology.Accordingly,this article aims to improve the equivalent source intensity solution method and the dual measurement surface sound field separation technology based on the equivalent source method.First,by establishing an equivalent source method near-field acoustic holography to solve the inverse problem of sound source identification,the current common norm constraints and their representative algorithms are introduced,namely Tikhonov regularization based on l2 norm constraints,and l1 norm sparse regularization.The optimized convex optimization solution toolbox(l1-CVX)and wideband acoustic holography(WBH)have carried out comparative analysis of reconstruction performance simultaneously,and summarized the advantages and disadvantages of each algorithm.Secondly,in order to solve the problem of poor recognition performance of coherent sound sources at low frequencies by MTwIST,the fast iterative shrinkage threshold algorithm(FISTA)under the sparse regularization framework is introduced to reconstruct the sound field,and FISTA is combined with high-order matrix function beam forming to enhance the sound.The dynamic display range of the source imaging results.Numerical simulations and experiments on FISTA,MTwIST,WBH,and Tikhonov regularized single sound source and coherent sound sources have verified that FISTA can reconstruct sparse sound sources with high accuracy in a wide frequency range,which is significant compared to other algorithms in the middle and low frequency bands.The advantage overcomes the limitation that the MTwIST threshold parameter is not suitable for lowfrequency coherent sound sources.FISTA’s high-end form FISTA-v can significantly improve the accuracy of sound source localization.Finally,the equivalent source grid arrangement for the traditional dual-measuring surface sound field separation technology cannot reflect the sparse characteristics of the sound source,leading to the problem of low separation accuracy.An improved sound field separation method is proposed.The measurement data is used to locate the actual position of the sound source,and the position information is used as a priori knowledge.The multi-island genetic algorithm is used to optimize the equivalent source grid layout to improve the separation accuracy.The results of reconstruction experiments on the separated sound pressure of the target sound source verify the effectiveness of the proposed improved method.