A Fault Diagnosis Analysis Of Power Transformer Based On Genetic Algorithm

Power transformer is the important unit of execution in the power system, power to change the stable operation of electrical appliances for the stable operation of the power system to play a crucial role, and the power transformer failure are often of a serious electrical equipment accident, but the result will be serious consequences, it is necessary for the potential failure of the power transformer diagnosis, early detection of problems in order to resolve.Nonlinear relationship because of the potential failure of the power transformer and transformer oil dissolved gas content, so the existing techniques are for the dissolved gas analysis and thus to infer the potential failure of the power transformer. However, traditional methods are prone to prematurely fall into local optimal solution to pick out caused by the fault diagnostic errors, in order to solve this problem, to seek the global optimal solution, this paper completed the following work:Genetic algorithm basic idea of the basic principles and basic concepts of a brief introduction and implementation of the necessary steps of the genetic algorithm and the operator is described in detail, and the genetic algorithm is introduced to pattern classification, a use of genetic algorithm to find the mathematical expression of the discriminant function; turn the genetic algorithm is introduced to the field of data mining, a genetic algorithm for decision tree last established under this article two types of algorithm and its application in power transformer fault diagnosis. Established in this paper two algorithms, respectively, relative to traditional algorithms, comparative results are:the change in electrical fault diagnosis based on genetic algorithm and discriminant function compared with the results of the traditional IEC three-ratio method, the total recognition rate from68.4%to90.1%; based on the results obtained by the genetic algorithm and linear decision tree of the power transformer failure diagnosis with traditional C4.5method compared to the total recognition rate from84.8%to90.1, the results show that genetic algorithm is introduced to the power change The electrical field of fault detection, fault diagnosis recognition rate.