Application of neural network with genetic algorithm to UHFPD pattern recognition in transformers

In this paper, an automated recognition system of ultra-high-frequency (UHF) PD designed by authors has been put forward to study the discharge properties in transformers. This paper presents Genetic Algorithm (GA) to train neural network (NN). Using BP-NN and GA-NN, we distinguish between basic types of defects appearing in transformers, such as corona, void, bubble, creeping discharge and floating discharge. Tests in laboratory give satisfactory results of classification. Compared with BP-NN, GA-NN can overcome slow convergence and possibility of being trapped at local minimum value. Thus, the convergence, discrimination and generalization ability of GA-NN is improved remarkably.