Abstract
A novel time-domain nearfield acoustic beamformer ( NABF ) is proposed to identify locations and strengths of noise sources. This technique is applicable to not only narrowband and broadband noise source but also stationary and non-stationary noise sources. Multichannel inverse filters are designed using least-square optimization in conjunction with Tikhonov regularization to mitigate the ill-posedness inherent in the underdetermined model-matching problem. Factors including the array aperture, microphone spacing, focus point spacing and distance of projection have major impact on the resulting source image. The distance of reconstruction (DOR) is selected according to the condition number of the propagation matrix. Boundary defocusing problem can also be alleviated with the aid of windowing design. Beam patterns are calculated for the inverse filters with windowing. Furthermore, a retreated focus point method is devised to prevent the singularity problem. The computation efficiency of the multi-channel inverse filtering is considerably enhanced by using a state-space minimal realization technique. As indicated by simulation results, the NABF technique proves effective in identifying noise sources by using two-dimensional (2D) arrays.