Discrimination Research of Power Transformer Magnetizing Inrush Current Based on Artificial Neural Network

In this paper, the originating principles and characters of magnetizing inrush currents, and the influencing factors are analyzed. The physical characters of magnetizing inrush currents and internal fault currents are compared through digital simulation under several conditions. The model of RBF NN is established by means of ANN toolbox in Matlab software. The sample data of magnetizing inrush currents and internal fault currents have been trained, and the performance of the trained ANN has been tested. The results showed that RBF NN can distinguish magnetizing inrush currents from internal fault currents correctly.