Two-Dimensional Dynamic Grid Compressive Beamforming Technology For Acoustic Source Identification

Microphone array based beamforming acoustic source identification technology has been widely used in the field of automobile,aerospace and high-speed train due to its advantages such as convenience,high efficiency,wide recognition range,accurate estimation in mid-high frequency,and available for medium-and-long measurement distance.With the increasing requirements for the accuracy of noise source identification in practical applications,various types of high-resolution beamforming algorithms have become research focus.Compressed sensing theory based compressive beamforming has attracted much attention due to its both high precision and high efficiency.Typical compressive beamforming algorithms are analyzed in this paper.The problem of performance degradation due to basis mismatch is explored and a new compressive beamforming method is proposed to overcome this drawback.Firstly,the delay-and-sum beamforming and the cross-spectral imaging function are derived based on the spherical wave assumption and the performance evaluation indexes are expounded.The conventional beamforming is characterized by low spatial resolution and severe sidelobe contamination.Therefore,three beamforming clearness methods including the deconvolution approach for the mapping of acoustic sources are given.The simulation results of acoustic source identification show that the clearness methods can significantly shrink the main lobe width and reduce the sidelobe levels,but there still exist shortcomings with the above clearness methods.Furthermore,an observation model of far-field plane wave measurement using a planar array is established.Based on this,the conventional fixed-grid compressive beamforming theory is given and two sparse recovery algorithms,namely,iterative reweighted l₁ minimization and orthogonal matching pursuit are derived.The simulation results of fixed-grid compressive beamforming show advantages of high resolution,high computational efficiency,low sidelobe levels,and strong robustness to interference.However,its performance deteriorates when sources do not coincide with the grids,namely that the basis mismatch occurs.To solve this issue,a two-dimensional dynamic grid compressive beamforming is developed for planar microphone array.The proposed method optimizes the objective function by iteratively decreasing the surrogate function based on the majorization-minimization framework,which leads to a gradual process to refine both the dynamic grid coordinates and the corresponding source strength distribution.The results of numerical simulation and experiment demonstrate that the proposed method can circumvent the basis mismatch and thus ensure higher location and quantification accuracy,comparing to the fix-grid approach.It can be applied to a planar array with microphones randomly distributing and does not require prior knowledge of signal-to-noise ratio or source sparsity.The dynamic grid compressive beamforming can provide high-resolution and low contamination imaging,allowing accurate estimation of two-dimensional DOAs and quantification of source strengths,even with a small number of microphones.