Hybrid microphone array signal processing approach for faulty wheel identification and ground impedance estimation in wheel/rail system

Abstract

This study investigates a microphone array signal processing approach in a theoretical and experimental manner for faulty wheel identification and localisation and ground impedance estimation in a wheel/rail system. The sound source location during wheel rotation is estimated by a broadband weighted multiple signal classification (BW-MUSIC) method, while the impedance of the corresponding ground surface is estimated by the Levenberg–Marquardt and Crank Nicolson (LM–CN) method. The accurate location of the faulty wheel is determined by the kurtosis beamformer. As an acoustics-based noncontact diagnosis method, this technique overcomes the challenge presented by contact between the sensors and the examined structures, and it is more applicable for impulsive signals with broadband features, such as impact noise generated from faults on the wheel surface. The accuracy of the location and impedance estimations is also examined in this study, and the results obtained from the numerical and experimental analyses are observed to be in good agreement. With multiple sound sources and the interference of ground reflection, the BW-MUSIC method can provide a separate and distinct localisation result in the sound map, while the LM–CN method can provide a preferable estimation result of the ground impedance, allowing the faulty wheel to be detected accurately.