Sound Source Identification Method Based on Location Optimization of Equivalent Sources

Abstract

The equivalent source method (ESM) based on nearfield acoustic holography (NAH) is an important technique for sound source identification, and the proper arrangement of the equivalent sources is critical to improve the accuracy of the identification results. Traditional approaches are hard to use for accurately locating low and medium frequency sound sources, therefore, cannot provide reasonable information for the collocation of the equivalent sources. To overcome the issue, a joint iterative regularized sound source localization method based on a combination of constrained L1 and L2 regularization is developed first, followed by the optimal placement of the equivalent sources near the localized sound sources. The sparsity and robustness of results are guaranteed by using the compound regularizers of Li and L2. The simulation results show that the traditional methods have a resolution of roughly 1/2 wavelength, whereas the proposed method has a resolution of just 1/3 wavelength. Two typical cases are analyzed in the measured experiment. The sound source amplitude reconstruction error dropped from 16.67% to 4.17% and from 20.83% to 11.46%, respectively. According to the results of simulation and indoor measurements, the method provides superior spatial resolution, source intensity quantification accuracy, and robustness to traditional methods.