Doppler Distortion Removal in Wayside Circular Microphone Array Signals.

Condition monitoring and fault diagnosis (CMFD) of key moving sources has attracted more and more attentions in recent years. Among many CMFD strategies, the wayside acoustic signal-based method can detect slight failure efficiently in time with a low-cost system. However, the nonlinear time-varying distance between the moving source and the microphone array would bring in serious Doppler distortion in the acquired signals. The Doppler distortion would result in serious amplitude distortion, frequency shift, and band extension, which is an obstacle for accurate CMFD. Considering the superior time-spatial characteristics of circular microphone array, this paper presents a short-time sparse singular value decomposition (SSVD) method based on wayside circular microphone array signals to eliminate the existed Doppler distortion. The procedure of the presented method is summarized as follows. First, the Doppler distortion array signals are decomposed into a series of array segments with a sliding window. Afterward, the corresponding moving angles (MAs) of each array segments are calculated with the introduced SSVD method. Finally, the fitting time-varying MA curve and time-domain interpolation resampling are employed to eliminate the Doppler distortion. Simulation results validated that the proposed method could achieve better localization accuracy under heavy background noise as compared to other traditional methods. Moreover, the effectiveness of the proposed method has been further verified with real experiments, which indicates excellent potential application prospects for moving vehicles, trains, planes, and so on. [ABSTRACT FROM AUTHOR]