Transformer Condition Evaluation Based On Multi-information Fusion

With the constant promotion of the condition based maintenance and the application of on-line monitoring in the power system, it can get more and more condition information, it has been important subject that how to apply transformer condition information efficiently and scientifically to judge the transformer's state. Then, the dissertation systematically analyzes the transformer's faults and fault detection technique and sort the transformer condition information; In order to adjust to the demanding of the information processing, it uses RON method to design RBF neural network; Using Neural network ensemble model to combine neural network with evidence theory, it sets up the model of transformer state evaluation based on multi-information fusion. Having considered the situation of on-line monitoring, the dissertation selected the necessary status information, and proved that the method has good veracity and reliability of diagnosis.