An approach to fault diagnosis for non-linear system based on fuzzy cluster analysis

An approach to fault diagnosis based on fuzzy clustering is proposed. First, the fuzzy model representing each state of the system is built by extracting fuzzy rules from the sample data using fuzzy clustering algorithm. Then, the modified fuzzy models for fault diagnosis are obtained based on the original fuzzy models and constitute a whole rule-base. Furthermore, a strategy for fault diagnosis based on fuzzy clustering is presented to detect and locate faults in the system. Finally, some experimental results are shown to illustrate the effectiveness of the proposed approach.