A fault diagnosis method based on ANFIS and bearing fault diagnosis
Junhong Zhang, Wenpeng Ma, Liang Ma · 2014
An integrated method of fuzzy clustering, rough sets theory, and adaptive neuro-fuzzy inference system (ANFIS) for fault diagnosis was presented. Xie-Beni cluster-validity was introduced into fuzzy c-means clustering algorithm, and a combination of genetic algorithm and gradient descent approach was applied, to discretize the feature parameters and obtain the decision table. In order to make up for the shortcomings of ANFIS that the fuzzy rules are difficult to determine and there are many redundancies, rough sets theory was applied to reduce the decision table to acquire sensitive features and inference rules. According to the reduction, ANFIS was designed, and genetic algorithm was employed to train the network. Applying the method to rolling element bearing fault diagnosis and comparing with several other methods, the result indicates that, the proposed method which could reduce features, obtain rules effectively and reach up to a high precision is superior to the others.