Maximum Gpq–mean deconvolution for the impulsive fault feature enhancement of rolling bearing

Abstract

The bearing fault signal is easily obscured by background noise and random shocks in the initial stage. The maximum Gpq–mean deconvolution (MGD) method is proposed to address the challenge of extracting fault feature signals in the presence of impact interference. The use of a nonlinear activation function in MGD enhances the distribution characteristics of the filtered signal. The proposed method adopts a new sparse measurement method, which enhances the sparse measurement capability and solves the problem of the difficulty in extracting periodic fault signals under impact. The superiority of the method in rolling bearing diagnosis is demonstrated through simulation and experimental analyses. In comparison with traditional methods, such as minimum entropy deconvolution (MED), optimal minimum entropy deconvolution adjustment (OMEDA), and maximum correlated kurtosis deconvolution (MCKD), the proposed method in this paper significantly improves the ability of extracting bearing fault signals.
Keywords: Minimum Entropy Deconvolution; Kurtosis; Blind Deconvolution; Sparse Metric