In order to solve the problem that it is difficult to locate and identify power transformer faults quickly and accurately

a synthetic diagnosis method using multi-neural network and evidence theory for transformer fault diagnosis is presented. This method combines the strong nonlinear fitting ability of neural networks with the advantages of evidence theory in dealing with subjective and objective contradictions in uncertain reasoning problems. Various kinds of dissolved gas data in oil and conventional electrical test data are dealt by using neural network’s excellent abilities of learning

memory and recognition. Integrating data fusion methods

neural network’s preliminary results are diagnosed by evidence theory and the accurate diagnosis of transformer faults is realized. It has been shown by experiments that the accuracy rate of transformer fault diagnosis is up to 73%.