Power Transformer Fault Diagnosis Based on Integrated of Rough Set Theory and Neural Network

Abstract

In this paper, a rough set (RS) and neural network (NN) integrated algorithm based fault a gnosis for power transformers, using dissolved gas analysis (DGA) is proposed. This approach takes advantage of the knowledge reduction ability of rough set and good classified diagnosis ability of NN. Power transformer fault parameters are reduced by rough sets, then work as BP neural network's input vector. Neural network initial weights are set according to the confidence of reduction parameters. Simulation results show that the combination of rough sets with neural network has good diagnostic ability.