A new transformer protection based on the artificial neural network model
Mingjie Chen, Xin Zeng, GongHua Li, Jian Luo · World Automation Congress · 2008
Obtaining the transformer internal parameters exactly for physical model based transformer protection is difficult because of the complex electromagnetic relation of the transformer, exemplified by transformer protection method based on the loop equation principle. According to the approaching ability of the artificial neural network, using artificial neural network to approach the electromagnetic relation of the transformer, the artificial neural network model is constructed to substitute for transformer physical model, identify the inner parameters on-line. Transformer protection based on the artificial neural network model is realized after parameter identification. EMTP simulation results demonstrate that the proposed method can recognize internal faults within a half cycle of their occurrence, with apparent fault features and less threshold value. It can discriminate low-level internal faults, rising superior to the magnetizing inrush.