Power Transformer Fault Diagnosis Based on Fuzzy Integral Fusion

A new method to identify power transformer faults based on fuzzy integral fusion is presented. Firstly, the membership function of the value of three-ratio are presented and the data of the value of three ratios are processed by fuzzy method. Then the faults are identified by four different radial basis function (RBF) neural networks. Finally results of identification from four different neural networks are fused by fuzzy integral fusion and then form the final diagnosis. The results of simulation by presented method have been proved to be more accuracy than single neural network in identifying the practical transformer fault case