An adaptive sensitive frequency band selection method for empirical wavelet transform and its application in bearing fault diagnosis

Empirical wavelet transform (EWT) is an adaptive wavelet based analysis which can be used to extract useful amplitude modulated-frequency modulated (AM-FM) mono components from a bearing vibration signal. However, the pre-requisite segmentation method on the Fourier support of a signal without a rigorous theoretical principle has limited the application of EWT in bearing fault diagnosis. To overcome this difficulty, an adaptive frequency band selection technique utilizing the Harmonic Significance Index (HSI) and Particle Swarm Optimization (PSO) is proposed in this paper. In this approach, HSI is employed as the fitness value of PSO to quantify the fault information contained in various frequency bands, and the optimal parameters (i.e., the lower cutoff frequency and the bandwidth) of the most sensitive frequency band (i.e., the band having the largest HSI value) identified by the PSO are then used in EWT to bandpass filter the original signal to extract the fault related AM-FM mono component. Results from the case studies presented in this work confirm the effectiveness of the proposed technique for bearing fault diagnosis. The proposed technique is particularly useful for situations where a bearing defect signal is contaminated by strong non-Gaussian noise such as random impacts from other mechanical components. (C) 2018 Elsevier Ltd. All rights reserved.