Diagnosis of oil-insulated power apparatus by using neural network simulation

One diagnostic process for oil-insulated power apparatus is based on the analysis of the chemical composition of gases evolved by the insulating oil. Normally this can be done only by a human expert. A considerable amount of information on the relation between chemical components and the faulty part of the power apparatus has been accumulated. This paper describes a neural network system which can be applied to existing diagnostic methods to enable analysis even by inexperienced engineers.