Rolling element bearing fault diagnosis based on support vector machine

Rolling element bearings are widely used in industrial applications. This paper presents a fault diagnosis method for rolling element bearings based on support vector machine (SVM). Firstly, the features are extracted from the vibration signals by the five-level wavelet packet decomposition algorithm using db2 wavelet. Then, the principal component analysis (PCA) is performed for feature reduction. Secondly, the multiclass SVM as a classifier is used to diagnose the bearing faults. A grid-search method in combination with 10-fold cross-validation is applied to find the optimal parameters for the multiclass SVM model. To validate the proposed method, an experiment of fault diagnosis for rolling element bearings has been carried out. The results show that the proposed method has high accuracy for bearing fault diagnosis.