A Motor Fault Diagnosis Method Based on Immune Mechanism

In this paper, a framework of fault diagnosis system is proposed, which is based on negative selection algorithm and the immune network model. Firstly, train the detectors by immune tolerance, and then detect if faults appear. Diagnosis experiments show that the system in normal pattern and abnormal pattern can be reflected by the self set and the non-self set completely through clustering algorithm. So the accuracy of diagnosis is improved. In the course of diagnosis, multiple diagnosis is proposed to process the data. If the data can´t be recognized exactly, the abnormity degree is presented, which is the evidence for experts to make decision.