| Transformer is one of the important hinge equipments in the electric power system. It is quite important for ensuring the safety running of electric power system.The dissolved gas analysis technology of the transformer has become one of the most common and most important technologies of the transformer fault diagnosis and that can diagnosis the fault typeã€site severity and trend of development of the transformer according to the composition and content of gas. It can discover the latent faults which are not easy found by using electrical test.However, this technology exists some uncertainties in the process of judgment of some specific type of faults. Aiming at this problem, this paper choose to use D-S evidence theory to solve uncertainty problem of results of power transformer fault diagnosis. A key point of the evidence theory is the basic trust distribution function.Currently most basic trust distribution function are constructed by the data obtained to detect, or are given by experience of people.The outstanding advantage of Duval triangle is it retained some misjudged data due to fall out the ratio limit values provided by the IEC ratio method. Fuzzy comprehensive evaluation can be said to be a comprehensive evaluation of an effective multi-factor decision making method when things are influenced by a number of factors. So this paper proposes two methods of basic trust distribution function research, one is based on the Duval triangle method and the other one is based on the fuzzy comprehensive evaluation. The first method based on the theory of Duval triangle and uses the interpolation method of finite element three-node triangle to construct basic trust distribution function. The second method by constructing membership functions and weight use fuzzy comprehensive evaluation to construct the basic trust distribution function. Then the diagnosis results of the above two methods respectively compared with the results which based on three ratio method,and verified the effectiveness of the two methods. Finally, using the combination rules of D-S evidence theory to fusion the diagnosis results. The example calculations show that the diagnosis results of algorithm proposed in this paper conforming to the actual situation. It can accurately determine the fault type of transformer. |