Rolling Bearings Fault Diagnosis Method Based on EWT Approximate Entropy and FCM Clustering

Abstract

In this paper, a novel method for rolling bearings fault diagnosis based on empirical wavelet transform with approximate entropy and fuzzy c-means clustering is proposed. The method can identify different types of rolling bearing faults. Firstly, the original vibration signals of rolling bearing are decomposed by empirical wavelet transform to obtain several amplitude and frequency modulation components that have physical meaning. Then, with correlation analysis, the first three amplitude–frequency modulation components most relevant to the original vibration signal are selected and their approximate entropy is calculated as the eigenvector. Finally, the constructed eigenvector matrix is used as the feature for fuzzy c-means clustering to realize the fault diagnosis of rolling bearing. The experimental results show that compared with the fault diagnosis method based on empirical mode decomposition or approximate entropy, the proposed rolling bearing fault diagnosis method has a better fault recognition effect.