Using Acoustic Signal Analysis for IMs’ Bearing Faults Diagnosis in a Moving Subway Train

The subway is a widely used mode of public transportation around the world. Induction motors (IMs) in subway cars are essential to the traction system. High performance can lead to mechanical and electrical issues in these IMs over time. However, monitoring their condition is challenging due to excessive noise. Bearings play a critical role within these motors and are particularly prone to defects. To detect such faults, acoustic signals are widely utilized for their noncontact characteristics, affordability, and accessibility. This article presents a methodology for bearing fault (BF) diagnosis of IMs based on acoustic signal analysis, Wiener filter, empirical wavelet transform (EWT), and convolution neural network (CNN). In this study, four IMs used in a metro wagon were evaluated while the train was moving. According to the results, the method is capable of detecting BF with an accuracy of 98.21% under challenging circumstances.