Rolling bearing fault diagnosis based on Slice Energy Entropy Spectral Correlation Density-Continuous Hidden Markov Model

Abstract

Taking advantage of the cyclostationarity property of the vibration signal when fault arises in rolling bearing, the paper proposes a new fault diagnosis method of rolling bearing based on Slice Energy Entropy Spectral Correlation Density- Continuous Hidden Markov Model (SEESCD-CHMM). Firstly, the method of SEESCD is used to extract the feature of rolling bearing four states’ (normal, inner race fault, outer race fault and ball element fault) data to form the training feature vectors. Then the training feature vectors are used to train a CHMM and the optimal parameters of CHMM are obtained. At last, the SEESCD method is used to extract the test data to form the test feature vectors. The trained CHMM model is used to diagnose the test feature vectors and perfect diagnosis results are got which is 100% accurate. In the end, the advantages and the much higher accuracy of the proposed method is verified by comparing with other intelligent diagnosis methods.