An optimal filtering frequency band search method based on MZGWO in rolling bearings fault diagnosis

Abstract

Envelope spectrum analysis based on bandpass filtering is one of effective methods for fault diagnosis of rolling bearings. Finding the optimal filtering frequency band is the key to extract the fault information and also a major long-standing challenge. The difficulty of selecting the optimal filtering frequency band increases when the signal-to-noise ratio (SNR) is low. In order to solve this problem, the amplitude probability distribution statistical feature of the bearing vibration acceleration envelope signal is utilized as the fitness function in this paper. An improved grey wolf optimization algorithm named multi-zone grey wolf optimization (MZGWO) algorithm is proposed to find an optimal filtering frequency band. Combined with the bandpass filter construction method, the optimal bandpass filter is established to extract the hidden fault information in the original signal. The proposed method is analyzed and verified by the simulation signals of the bearing dynamics model and the experiment signals of the fault bearings, and compared with several optimal filtering frequency band search methods. The analysis results show that the proposed method is effective and has advantages in the search for the optimal filtering frequency band.