TQWT-assisted statistical process control method for condition monitoring and fault diagnosis of bearings in high-speed rail

Abstract

Condition monitoring and fault diagnosis of bearings in high-speed rail have attracted considerable attention in recent years, however, it’s still a hard work due to harsh environments with high speeds and high loads. A statistical condition monitoring and fault diagnosis method based on tunable Q-factor wavelet transform (TQWT) is developed in this study. The core idea of this method is that the TQWT can extract oscillatory behaviors of bearing faults. The vibration data under the normal condition are first decomposed by the TQWT into different wavelet coefficients. Two health indicators are then formulated by the dominant wavelet coefficients and the remaining coefficients for condition monitoring. The upper control limits are established using the one-sided confidence limit of the indicators by using the non-parametric bootstrap scheme. The Shewhart control charts on multiscale wavelet coefficients are constructed for fault diagnosis. We demonstrate the effectiveness of the proposed method by monitoring and diagnosing single and multiple railway axle bearing defects. Furthermore, the comparison studies show that the proposed method outperforms a traditional time-frequency method, the Wigner-Ville distribution method.