Research On Support Vector Machine And Its Application On Fault Diagnosis

Support vector machine (SVM) is a new learning method based on statistical learning theory, which has been successfully applied to many areas. Firstly, the basic theory and arithmetic was analyzed in detail. Secondly, the generator fault diagnosis model based on SVM and integrated characteristics was proposed. In order to get typical characteristics, the vibration signal and electrical current signal were chose respectively as features of the model. Then, the method of Least Squares Support Vector Machine (LS-SVM), which is the betterment of SVM was applied to fault diagnosis of rolling bearing. The energy of each node was regarded as eigenvector of diagnosis models and was input to LS-SVM multi-classifier to recognize failures using wavelet packet transforms. Finally, the validity of the above two methods were validated with corresponding experiment table.