POWER-PLANT FAULT-DIAGNOSIS USING A FUZZY KNOWLEDGE-BASED SYSTEM

This paper presents a methodology for power plant fault diagnosis, whi ch is based on inductive machine learning and fuzzy logic. Expert know ledge concerning normal and faulty operation is acquired via knowledge acquisition techniques. We aim toward the construction of a dynamical ly changing fuzzy rule-based system, able to diagnose plant anomalies. The proposed system, deals with two important malfunctions of the boi ler's circuit, namely, tube leakage and fouling. We demonstrate our ap proach by experimenting with real world data taken from a specific the rmal plant.