Fault Diagnosis of Rotating Machinery Using Data Mining and CLIPS-Based Expert System

Expert system is widely used in fault diagnosis, but how to obtain knowledge is a bottleneck restricting its development. Data contains rich information about the machine, and acquiring knowledge from it or data mining in other words is considered as an effective way solving this problem. An algorithm based on rough set and genetic algorithm for feature reduction is proposed in this paper. Minimal assemble of necessary features i.e. reduct is computed, from which diagnosis rules can be generated. On the base of the rules discovered, an expert system for fault diagnosis of rotating machinery is presented. It is a production rule-based expert system developed on CLIPS, and realized together with VC++ programming and SQL SERVER database. To validate the diagnosis system, an experiment is carried out in rotor test-bed and further improvement is pointed out.