Deconvolution algorithms of phased microphone arrays for the mapping of acoustic sources in an airframe test

Abstract

Beamforming methods with phased microphone arrays are widely used for the characterization of acoustic sources. Compared with conventional beamforming, deconvolution algorithms, such as DAMAS, NNLS, FISTA, and SpaRSA, can significantly improve the spatial resolution but require huge computational effort. However, there are no experimental applications of FISTA and SpaRSA for complex acoustic sources localization. In this paper, deconvolution algorithms (DAMAS, NNLS, FISTA, and SpaRSA) are compared through experimental applications of benchmark test DLR1, which contains complex sound sources distributions. Results show that DAMAS, NNLS, and FISTA can distinguish dominant acoustic sources for a broad range of frequencies in complex acoustic sources distributions cases. SpaRSA has great possibilities of being invalid in complex aeroacoustic measurements. Besides, DAMAS, NNLS, and FISTA with compression computational grid based on conventional beamforming (denoted as DAMAS-CG2, NNLS-CG2, and FISTA-CG2, respectively) successfully save lots of computational time and retain the spatial resolution. Therefore, DAMAS-CG2, NNLS-CG2, and FISTA-CG2 can be considered as the promising acoustic array data processing methods for complex acoustic sources localization.