Expert system for power generation fault diagnosis using hierarchical meta-rules

Rule processing and selection in forward chaining is a critical factor in the performance of expert systems. Considered the large quantity of information and rules in the hydroelectric generator unit, the rules were organized using hierarchical meta-rules. Meta-rules have been used to improve rule selection and increase the speed processing. This paper describes a method for improving rule selection automatically according with the device condition. The expert system developed in JESS (Java Expert System Shell) processes rules using rete algorithm. The FMEA (Failure Mode and Effect Analyze) is used as a tool to perform rules, that was implemented in CLIP language. The information required to perform FMEA is collected from past Work Order (WO), Maintenance Technical Instructions (MTI), Operation Technical Instructions (OTI) and Autonomous Maintenance Plan (AMP). This expert system assesses faults on power generation systems, subsystems and equipments. The diagnosis system processes online information acquires from fieldbus sensors and transmitters. This diagnosis method implemented can provide warning of failure; reduce the inspection cost, downtime and inventory. This expert system has been applied to hydroelectric power plant in Brazil. The results obtained along with their evaluations are satisfactory.