Incipient Fault Feature Extraction of Rolling Bearing Based on Full Vector Complete Ensemble Empirical Mode Decomposition

Abstract

Feature extraction plays a crucial role in the rolling bearing fault diagnosis, which determines the results of the fault diagnosis largely. However, conventional fault diagnosis methods only use single channel signal for analysis, which may lead to missed or misjudged diagnostic results. This work presents an incipient fault feature extraction based on full vector complete ensemble empirical mode decomposition (FVCEEMD) to extract the fault feature of rolling bearing. Complete ensemble empirical mode decomposition (CEEMD) is introduced as a pretreatment method to decompose signals into a set of intrinsic mode function (IMFs). Parameter index for selecting sensitive IMF components is correlation coefficient and kurtosis. And hilbert transform is used to obtain corresponding envelope signals. The full vector spectrum is utilized to analyze the envelope signals to get fault characteristics of rolling bearing. The results have been verified that the presented method can effectively extract the incipient fault feature of rolling bearing, and it has broad application prospects in industry.