Rolling Bearing Fault Diagnosis Based on Wavelet Package Transform and IPSO Optimized SVM

An intelligent rolling bearing fault diagnosis method based on wavelet package transform (WPT) and an improved particle swarm optimization algorithm (IPSO) optimized support vector machine (SVM) is proposed. Firstly, WPT is adopted to obtain the energy distribution of the collected signals and the feature vectors are selected and extracted based on the decomposed different frequency signal components. Secondly, the parameter optimization process of SVM is conducted based on the proposed IPSO to improve the diagnosis accuracy. The improvements of IPSO are achieved by using the chaotic tent map and Lévy flight mechanism, where the chaotic tent map is used to initialize the solution space and the Lévy flight mechanism is selected to maintain the diversity of the population. Finally, feature vectors of different fault types are inputted into IPSO-SVM to classify different fault samples. Based on the well-known CWRU bearing dataset, extensive comparison tests are carried out to show the effectiveness and superiorities of the proposed method.