Doppler effect removal based on short-time sparse SVD strategy for wayside acoustic source monitoring

摘要

As wayside condition monitoring based on acoustic signal shows the merits of low cost and high efficiency, it has become more and more popular in recent years. However, the serious Doppler distortion in the obtained signal of the key acoustic source would increase the difficulty for accurate condition monitoring. This paper proposed a novel method based on short-time sparse SVD (ST-SSVD) to remove the Doppler distortion using a linear microphone array. First of all, the acquired array signal is divided into a series of array se gments for calculating the time-varying direction of arrival (DOA) of the target source. Afterwards, time interpolation resampling (TIR) is employed to remove the Doppler distortion with the established resampling time series, which is calculated according to the estimated time-varying DOAs. The case study has validated that super localization accuracy and better robustness to noise could be achieved as compared to other traditional strategies like ST-MUSIC. The proposed method shows great application value for the wayside condition monitoring system for moving vehicles, trains, planes and etc.