An optimized nearest prototype classifier for power plant fault diagnosis using hybrid particle swarm optimization algorithm

Correct and rapid fault diagnosis is of great importance for the safe and reliable operation of a large-scale power plant. It is a difficult task, however, due to the structural complexity of a power plant, which needs to deal with hundreds of variables simultaneously in case of fault occurrence. A novel nearest prototype classifier is proposed in this paper to diagnose faults in a power plant. A constructive approach is employed to automatically determine the most appropriate number of prototypes per class, while a hybrid particle swarm optimization (HGLPSO) algorithm is used to optimize the position of the prototypes. The aim is to generate an automatic process for obtaining the number and position of prototypes in the nearest prototype classifier with high classification accuracy and low size. The effectiveness of the HGLPSO classifier is evaluated on eight real world classification problems. Finally, the classifier is applied to diagnose faults of a high-pressure feedwater heater system of a 600-MW coal-fired power unit. The obtained results demonstrate the validity of the proposed approach.