Application of neural network and DS evidence fusion algorithm in power transformer fault diagnosis

As the transformer´s fault types and fault positions are complexity, complementarity, redundancy and strong characteristics of uncertainty, a satisfactory diagnosis of transformer fault types and fault position may not be obtained if only one technique is used. In this paper, a synthetic diagnosis method using neural network and DS evidence theory for transformer fault diagnosis is presented, combining DGA data with application of data fusion theory. This method has advantage of both neural and DS evidential theory, it can effectively solve the problem of uncertainty, and improve the fault diagnosis system´s accuracy and reliability. A simulation example in this paper illustrates that the method combing of neural network with DS evidence theory is indeed greatly improved integration of the credibility of the data, making diagnostic systems easy to design, high precision and prone to operation. It can fulfill users´ requirement perfectly.