An integration of neural networks and nonmonotonic reasoning for power system diagnosis

Presents a hybrid AI system, integrating neural networks and nonmonotonic reasoning, to be used as an operator's aid in the diagnosis of faults in power systems and in their training. Once the faults are localized by the neural network, the nonmonotonic reasoning subsystem analyzes the results and gives an explanation for them. The hybrid system can handle single, novel, noisy and multiple faults. The authors present in detail a case example of a simplified power system generation plant. The results obtained demonstrate that this hybrid system is a very powerful and reliable method for the solution of existing problems in power system diagnosis.