19th Asia-Pacific Vibration Conference (APVC2021)

280 / 2021-04-15 23:27:10

Application of a Novel Spectral Kurtosis Method for Railway Axle Bearing Fault Diagnosis

Fault Diagnosis, Railway Axle Bearing, L-kurtosis, Spectral Kurtosis, Kurtogram.

Under the interferences of harmonics, strong random impulses and strong noise, the high-frequency repetitive impulses induced by local defects in rolling bearings are easily submerged in the measured vibration signal. Thus, accurately identifying the informative frequency band containing high-frequency resonance in the bearing vibration signal is vital for diagnosing bearing faults. An improved spectral kurtosis method for railway axle bearing fault diagnosis is presented. The proposed method adopts L-kurtosis of power spectrum amplitude of envelope of frequency-band signal obtained by a 1/3-binary tree filter bank as an indicator to determine the optimum frequency band for demodulation. The power spectrum of demodulation signal with maximal L-kurtosis value is used further to identify bearing fault types. The effectiveness and robustness of the proposed method for bearing fault diagnosis is validated by using the experimental data collected from railway axle bearing test bed.