Application of Rough Sets, Neural Network and Expert System to Power System Fault Diagnosis

Wu DENG,Xin-hua YANG,Hui-min ZHAO,Fei-long TANG

2009-01-01

Abstract:In accordance with characteristics of more indeterminate information and higher speed request in power system substation fault diagnosis system, on the basis of switch and relay protecting information of substation, according to the intelligence complementary strategy, a new substation fault diagnosis method based on rough sets-neural network-expert system was presented. Firstly, based on data acquisition and pretreatment, the original fault diagnosis samples were discretized by the hybrid clustering method. Then, the decision attribute was reduced to delete redundant information for obtaining the minimum fault feature subset. In the course of identifying fault diagnosis through radial basis function (RBF) neural network, some output results of RBF neural network was modified by using the inference capability expert system. The results show that the presented method is effective by applying the presented method to the certain substation.