An Adaptive Frequency Window Empirical Wavelet Transform Method for Fault Diagnosis of Wheelset Bearing

Abstract

Aiming at the problem that it is different to extract weak fault feature for wheelset bearing under strong background noise, a novel method called self-adaptive frequency window EWT (empirical wavelet transform) is proposed. Firstly, a moving and flexible frequency window is introduced to segment the Fourier spectrum of signal, which is based on the Fourier transform of fault signal of bearing. Secondly, an evaluating index combining envelope spectrum kurtosis (ESK) and envelope spectrum harmonic-to-noise ratio (ESHNR) is used to determine the position of frequency window through the water cycle algorithm (WCA) optimization method. Finally, the bearing fault features are extracted by envelope demodulation analysis of the optimal mode signal generated by the frequency window EWT. The result shows that this method can effectively improve the weak fault detection of bearing and eliminate the interference of the strong background noise.