Multi-Source Signal Localization Analysis via ST-IMUSIC for Trackside Acoustic System
Ying Zhang, Yongxiang Zhang, LI Pei-ran, Jinchang Cheng, Wenhao Cheng, Xiaoxi Ding · 2023
Bearing defects are the main cause of train failures. Diagnosis by trackside acoustic system has the advantages of low cost, non-contact measurement and sensitivity to early faults, but there are difficulties in estimating the location of bearing failures accurately. Therefore, on the basis of classical multiple signal classification, this paper proposes an improved short time sliding window method, called ST -IMUSIC (Short time-improved multiple signal classification). The method consists of three parts: Firstly, divide array data into a series of short-time segments by sliding windows to meet the application conditions of classical MUSIC algorithm; Secondly, reconstruct array data and covariance matrix using the short-time signal segments, and then estimate the spatial spectrum of them according to the MUSIC steps; Finally, search the spatial spectrum peaks of each short-time segment to obtain the time-varying positions of the trackside acoustic multiple source signals. Besides simulation analysis, this paper also builds a trackside acoustic array sensing system based on a MEMS miniature digital microphone and an embedded development board, and verifies the proposed method through dynamically collected experimental signals. The results show that the system can achieve simultaneous multi-channel data acquisition and transmission, and ST-IMUSIC can realize trackside acoustic multi-source localization.