Periodic component pursuit-based kurtosis deconvolution and its application in roller bearing compound fault diagnosis

Abstract

As an effective roller bearing fault diagnosis method, Adaptive Periodic Mode Decomposition (APMD) method has excellent capability of repeated transient extraction. In APMD, the Maximum Likelihood Estimation (MLE) method is used to calculate the projection energy, and the period corresponding to the maximum projection energy is taken as the main period of the present signal. When the noise is large, the noise energy will greatly interfere with the projection energy, thereby affecting the accuracy of the period estimation. To solve this problem, a Periodic Component Pursuit-based Kurtosis Deconvolution (PCPKD) method is proposed, which uses Maximum Reweighted Kurtosis Deconvolution (MRKD) to de-noise the signals that need to estimate the main periods, so as to reduce the influence of noise energy on the periodic estimation to the greatest extent. The proposed method is applied to roller bearing compound fault diagnosis, and the experimental results show that PCPKD has a good ability of signal period enhancement and can effectively segment multi frequency components.