Research On Fault Prediction And Condition Assessment Of Power Transformer

Inorder to guarantee the security,reliability and economical efficiency of power grid,power enterprise have concerned with the fault prediction and condition maintenance of electric equipments,which can help them reduce fault chance rate and make rational maintenance strategy.Aiming at the dissolved gas in power transformer, we present a combination forecasting model with changing weight value based on support vector manchine and then propose a fault classifier based on Bayesian network.The two models work together and then make fualt prediction.The other outcome of our research is condition maintenance. In this paper, transformer is devided into three parts: main body,bushing and cores,we will assess all of the parts individually.With a hierarchical evaluation model based on 5-rank conditon and centensimal system, we could obtain running conditions in different periods including previous,present and prospective of the transformer.Then we send the conditions into an evaluation model based on Bayesian net work and fianlly draw the synthetic conclusion, which can help us make decision support on condition maintenance.