Artificial Neural Networks Busbar Protection

Bus-bar is an very important part in Power System, when an fault happens on the bus-bar, all the components connected to the bus will have power failures. Before the bus-bar returning to normal or switching to another bus-bar, all these parts can't work..The main bus-bar protection methods at home and abroad are LXB type, IC type, computer type and so on, However, these are all defective.So the study on bus-bar protection is of great significance.Artificial neural network, as the rising new technical disciplines in recent years, have played a huge role in medicine, biology, finance, petroleum, chemical and other industries. Artificial neural network also plays a great role in the study of Power system protection.The paper first introduces some characteristics of the power system protection, analyzes several bus-bar protection methods judging from power system protection, chooses the busbar differential protection as the core principle. Secondly, the paper introduces some of the concepts, characteristics and structure of neural networks, besides, some introduction of BP neural network and RBF neural network.The paper has studies on current transformer, when excluding the non-periodic component or containing non-periodic componentsn, makes research on the saturation of the current transformer and discusses the impact of the saturation characteristic on bus-bar protection.The paper also analyzes the bus-bar model based on classification and approach ability, using BP neural network and RBF neural network, reach the target of the classification of the normal operation,the fault outside the bas-bar, the fault inside the bas-bar, besides,we can have more refined variety of the type of fault.