Bearings Fault Diagnosis Based on Multiwavelet Energy Statistics Measure

In order to solve the problems of correctly identifying incipient fault for bearings and improve classification ability, the new scheme for bearing fault diagnosis based on multiwavelet energy statistics was proposed. The signal energy spectrum in multiwavelet domain was used as fault diagnosis characteristics. With the distance evaluation technique, the optimal features sub-filed were obtained. The optimal features were input into the SVM to identify the different fault cases. The Receiver Operating Characteristic curve (ROC) was applied to evaluate the effect of different multiwavelets preprocess methods. Finally, the experimental results show that the proposed methods can more efficiently opposes the characters of different fault cases and diagnose bearings faults with the appropriate preprocess methods.