Research on bearing fault diagnosis using APSO-SVM method

According to the statistics, over 30 % of rotating equipment faults occurred in bearings. Therefore, the fault diagnosis of bearing has a great significance. To achieve effective bearing faults diagnosis, a diagnosis model based on support vector machine (SVM) and accelerated particle swarm optimization (APSO) for bearing fault diagnosis is proposed. Firstly, empirical mode decomposition (EMD) is adopted to decompose the fault signal into sum of several intrinsic mode function (IMF). Then, the feature vectors for bearing fault diagnosis are obtained from the IMF energy. Finally, the fault mode is identified by SVM model which is optimized by APSO. The experiment results show that the proposed diagnosis method can identify the bearing fault type effectively. © 2014 IFSA Publishing, S. L.