Sample entropy based roller bearing fault diagnosis method
Zhao Zhi-hong1,2; Yang Shao-pu2
1. School of Mechanical, Electronic and Control Engineering, Beijing Jiaotong University, Beijing 100044, China;2. School of Computing and Informatics, Shijiazhuang Tiedao University, Shijiazhuang, 050043, China
Abstract:The nonlinear dyanmics parameter sample entropy is used as feature for roller bearing fault diagnosis. Normal, inner race, ball, outer race fault bearing were used for analysis and diagnosis. The sample entropy of the original vibration signal calculates the entropy at one scale. Information about the characteristics of the vibration signal at different scales can give important information about the fault. A sample entropy method that based on the Ensemble Empirical Mode Decomposition (EEMD) is proposed in this paper. First, the original roller bearing vibration signal is decomposed by EEMD and the intrinsic mode functions that contained the most information are chosen to calculate the sample entropy and selected as feature vector. SVM method was used as classifier to identify different faults. Thus the vibration signal analysis in different scales can give more information about the fault. Experimental results with real roller bearing data show that the proposed method is effective.