Fault Diagnosis Method of Power Transformer Based on Improved PNN

Abstract

Abstract Power transformer is a core hub of power transmission in power grid, and its operation state will affect the effective operation of power grid. The existing power transformer fault diagnosis methods mainly use the traditional dissolved gas analysis (DGA) method combined with artificial neural network. Probabilistic neural network (PNN) has a wide range of applications in the field of power transformer fault diagnosis because of its good fault tolerance and high training efficiency. In PNN, the smoothing factor has a great influence on the output, but the parameter is lack of efficient selection method, which makes the classification efficiency of PNN not high. Aiming at this problem, a diagnosis method based on improved PNN is proposed for power transformer faults in this paper. Bat algorithm is used to optimize the smoothing factor of PNN to determine the optimal value. Based on the optimal results, the network model is trained to obtain the optimal power transformer fault diagnosis model. The experiment fully proves that the improved PNN has higher diagnosis accuracy than other methods.