A Method of State Diagnosis for Rolling Bearing Using Support Vector Machine and BP Neural Network

摘要

By utilizing the SVM and neural BP network, a method of state diagnosis for rolling bearing is presented. The SVM is used to establish a classifier for the normal and fault state, then two kinds of samples caused by the distinct states are trained to judge whether the rolling bearing is normal or false. If the rolling bearing is in the fault state, all the fault samples were trained by the classifier composed of BP neural network to recognize which fault state it is in, otherwise, the state diagnosis is finished. The final experiment results show that the proposed method can diagnose the fault type more quickly and effectively in the small sample circumstances compared with the one using the BP neural networks solely.