Yang Deyou

The transformer fault diagnosis based on Quantum Neural Network

It is very difficult to analyze the fault for dissolved gas analysis(DGA) because of the less and uncertain transformer gas information. In this paper, the Quantum Neural Network(QNN) was applied to diagnosis the transformer fault by employing the DGA data. The QNN can be used to reduce the uncertainty of pattern recognition by allocatting the uncertain data to the right fault pattern quickly and rationally as its model draws on the ideas of quantum phase computing and the complementary amendment relationship exists between the hidden layer output and phase shift parameters. The validity and feasibility of the proposed method were verified by testing the real DGA data. The results of the propoed model was compared with BP neural network in dealing with the actual DGA data also.

A new method based on fuzzy TOPSIS for transformer dissolved gas analysis

This paper presents a new classifier based on Fuzzy Technique of Order Preference by Sunilarity to Ideal Solution (TOPSIS) for power transformer fault diagnosis in dealing with the dissolved gas analysis(DGA) problem, which is treated as the collision between lowering complex process of the analysis and increasing quantity of monitoring informations. Two methods are employed to ameliorate the diagnosis condition. Firstly, the essential concept of Fuzzy Set using semantic variables is introduced. And this classification approach is then combined with the TOPSIS considering the efficiency in managing imperfect and incorrect datas, and the multiple attribute decision matrix is deduced by fuzzy mapping. Secondly, the vector interval is calculated, which fixes the relative similarity of the possible fault types on the evidence of various gas ratios. Finally, contrasted with IEC Three-ratio method, the new method has been proved rapid and exact in judging the operation states, and therefore the safety, stability and economy of transformer are improved.