Study Of Near-field Acoustic Holography Algorithm Based On Sparse Equivalent Source Method

The identification and location of the noise source is the important prerequisite for noise control.In recent years,with the rapid development of computer technology,near-field acoustic holography technology based on acoustic array measurement has been continuously and extensively researched and gradually applied to automobiles and aerospace and other fields.The near-field acoustic holography technology based on the equivalent source method has received extensive attention because of its suitability for arbitrary-shaped sound sources,good robustness,high efficiency,and easy implementation.The core of the equivalent source method is the solution of the source intensity vector,and different regularization solutions will produce different recognition effects.When the sound source is sparse or close to sparse,the advantages of the sparse equivalent source method are highlighted.Based on the sparse equivalent source method,this paper aims to explore higher performance sound source recognition methods with reconstruction performance and sound source recognition resolution.Firstly,the basic principle of the equivalent source method is explained and three classic equivalent source intensity solving algorithms are introduced,namely Tikhonov regularization algorithm,wideband acoustic holography（WBH）and fast iterative shrinkage threshold algorithm（FISTA）.The three algorithms are simulated and analyzed,and the shortcomings of the three algorithms are explored,and the advantages of the sparse regularization method in sound field reconstruction are pointed out.Based on this,this paper proposes an augmented Lagrangian sound source recognition method based on fixed point iteration（ALM-FP）under the framework of compressed sensing to achieve better sound field reconstruction performance over a wider frequency band.The simulation and experimental comparative analysis of single sound source and double coherent sound source are carried out.The results show that compared with the two sparse regularization methods of WBH and FISTA,ALM-FP can achieve better reconstruction accuracy in the entire analysis frequency band,and the resolution of sound source recognition is higher.However,the reconstruction accuracy of the algorithm will decrease when the frequency decreases,and there is still a certain gap between the reconstructed sound pressure and the theoretical results in the low frequency band,which can be further improved.In order to further improve the sound field reconstruction performance and resolution of sparse sound sources in the low frequency range,this paper proposes an equivalent source method near-field acoustic holography based on_2/3-norm regularization(_2/3-norm).This method constrains the source intensity vector with the_2/3 norm,so that the source intensity vector has stronger sparsity in the inverse solution,which is an important reason for its good reconstruction performance in the low frequency band.Simulations and experiments of single and dual sound sources show that,compared with FISTA and ALM-FP,the _2/3-norm algorithm has the sparser source intensity distribution,and has the smaller reconstruction error in the middle and low frequency bands,and the sound source recognition resolution is higher.In addition,simulation shows that the method has good adaptability to the layout parameters such as signal-to-noise ratio,holographic distance and the number of microphones.