Compound Fault Diagnosis in Railway Vehicle Wheelset Bearing Based on ISAM-AHKD

Accurately diagnosing compound faults in wheelset bearings is a challenging task. This is mainly due to the fact that the vibration signals of the bearings are affected by wheel-rail interactions, and it is also difficult to extract from these signals the features of the corresponding compound faults. To overcome these problems, we propose a method consisting of improved spectral amplitude modulation (ISAM) and adaptive hyperbolic kernel distribution (AHKD): the ISAM-AHKD. ISAM employs weighted spectral trend (WST) and magnitude order (MO) as modulation operators for nonlinear filtering in the frequency domain; the new signal is obtained by inverse Fourier transform. This new signal is further introduced into a Bayesian-optimized AHKD based on Rényi entropy. ISAM has excellent fault feature enhancement capability and high robustness. It can reduce the effect of wheel-rail interactions on wheelset bearing signals. AHKD has high time-frequency resolution and no cross terms. It can effectively identify the compound fault features. Through simulated and experimental validation, the ISAM-AHKD is shown to be able to effectively detect compound faults in wheelset bearings in the presence of strong excitations generated by wheel-rail interaction. This result demonstrates its potential and usefulness in industrial applications.