Application of Entropy Information on Bearing Fault Diagnosis

Abstract

Zihan Wang and Yongjian Sun

In order to solve the problem of nonlinear, nonstationary, complex components and redundant information of rolling bearing vibration signal in single scale, a rolling bearing fault feature extraction method based on wavelet packet decomposition and permutation entropy and sample entropy is proposed. Firstly, wavelet packet decomposition is used to decompose the original signal of rolling bearing into several sub bands with different frequencies, and the permutation entropy and sample entropy of signal data at different frequencies are calculated. Secondly, the sample entropy and permutation entropy of different frequency signals after decomposition and reconstruction are extracted to form a high-dimensional feature vector to complete the initial fault feature extraction. Finally, the extracted feature samples are randomly arranged for fault recognition. The experimental data of rolling bearing processed by this method are identified by extreme learning machine. The results show that the method can effectively identify the fault types of rolling bearing, and the classification effect is better than that of the original data set training, and the classification accuracy reaches 99.8.

PDF