Station Alarm Message Dealing System Based ANN And ES

Many alarming messages will occur when there are some fault arising in substation, but in these messages some are unreliable. It is very difficult for the operator to decide where the fault happened under this condition. For these sakes it is a significant and essential task that substation alarm message dealing system is developed to assist operators. On the background, this paper studies the fault diagnosis using Expert System (ES) and Artificial Neural Network(ANN) because using ES or ANN separately has flaw. Firstly, the paper gives an brief account of ANN and researches the fault-tolerance performance of BP NN, it also puts forward a practical calculation which can improve the fault-tolerance performance. Secondly the paper develops the alarming dealing system directed at NanChang 5 00KV automation substation. The kernel components of this system are bus ,electrical line and transformer. It constructs some network-diagnosis model classified by the kind of the components whose input messages are the information of protection alarm and tripped breakers, then the paper builds up a network-diagnosis model of substation on this basis. The diagnosis kernel of this model is ANN with the fault-tolerance performance and it combines the ES. With the inference an judgment ability of ES the model can judge the operating way and modify some output of ANN.