The Identification Of Transformer Inrush Current Based On Wavelet Transformer And Probabilitistic Neural Network

As one of the most important and expensive electrical equipments in the power system, power transformer bears an important role of electricity transmission and voltage transform. Its reliable operation is directly related to whether the power system can work safely and steadily. Therefore, the protection research of power transformer has always been an important topic in power system protection. The identification of inrush current is the key problem in transformer differential protection, and it is of important significance in improving the accuracy of transformer protection tripping and the quality of power supply of power system. Hence, domestic and overseas scholars have done a lot of study work, and many magnetizing inrush current identification methods are proposed. However, various identification methods have their respective advantages and disadvantages, and they still cannot meet the requirements of modern microcomputer relay protection. So using the new principle and method to identify transformer magnetizing inrush current has the reality urgency.In this paper, the formation mechanism of three-phase transformer inrush current and its effect to differential protection is studied and analyzed. This paper establishes the models of three-phase transformer inrush current and internal fault current in Matlab/Simulink software, carries out the digital simulation, and compares the distinction between two kinds of currents. Discrete wavelet is selected to decompose three-phase transformer inrush current and internal fault current to multi-layer. The wavelet’s energy spectrum is obtained and characterizing quantity extracted. The paper presents the transformers differential protection scheme based on Probabilitistic Neural Network (PNN), and training samples and test samples are used to train and test the network. The best spread is obtained through much adjustment, and the simulation results show that this method can identify three-phase transformer inrush current and internal fault current with high accuracy and speed. In the end of the dissertation, Genetic Algorithm (GA) is used to obtain optimal smoothing factor for PNN, which raises the ability and accuracy of PNN.