EEMD Method and TWSVM for Fault Diagnosis of Roller Bearings

Abstract

The ensemble empirical mode decomposition (EEMD) is a self-adaptive signal processing technique for nonlinear and non-stationary signals, which can alleviate the mode mixing problem occurring in empirical mode decomposition (EMD). As a improved support vector machine (SVM) method, Twin support vector machine (TWSVM) is a powerful tool for supervised learning, which are successfully applied to classification and regression problems. This paper proposes an effective fault diagnosis method for roller bearings based on EEMD and TWSVM. First, the vibration signals collected from the roller bearings are decomposed using EEMD and intrinsic mode functions (IMF) are produced. Second, the sample entropy of the most IMFs are calculated as the feature of initial signal. At last, these features, as training and recognition samples, are fed into TWSVM to identify the bearing fault conditions. The experiment results show that the proposed method can accurately recognize the bearing normal, inner race, outer race and ball fault under small samples.