Rolling element bearing diagnostic based on EEMD and SVM

Abstract

Considering rolling bearing incipient fault feature is not obvious, with bearing vibration signals, a bearing-condition-monitoring method based on the ensemble empirical mode decomposition(EEMD) and Support Vector Machine(SVM), has been proposed in this paper. EEMD is used for adaptive decomposition of bearing vibration signals into several intrinsic mode functions (IMFs). The IMF component energy characteristics and dimensionless factor of time domain characteristics were extracted as the original features. After that, Back Propagation (BP) neural network and Genetic Algorithm (GA) were employed for feature selection. Then, dimension-reduced feature based on principal component analysis (PCA) is to establish the SVM model and monitor bearing fault state. The experiment shows that with this method diagnosis of rolling element bearing faults could be realized effectively.