Study On Fault Diagnosis Of Power Transformer Based On Multi-dimensional Characteristic Quantity

As the most critical high voltage electrical equipment in the power grid,the transformer plays a key role in the transmission,distribution and conversion of electrical energy.The unplanned outage caused by the transformer fault may cause the interruption of energy transmission,thus affecting the stability of the whole system and bringing huge economic losses.Therefore,with the state maintenance strategy getting in further promotion in power system,determining the state of the transformer fault,making accurate judgments on the transformer internal fault or abnormal state and developing the maintenance or repair strategies timely are all benefit to reducing the loss and damage due to abnormal transformer accident,which is of great significance to the whole system.Based on the problems and features existing in current transformer fault diagnosis method,this paper considers combining the transformer on-line and off-line data as fault characteristic parameters in fault diagnosis,so as to explore to build multidimensional volume based decision model.A fault diagnosis model based on rough set and multi class support vector machine is proposed.A detailed analysis of transformer fault factors,this paper considers dealing with the relevant information of the different kinds of fault in collaborative analysis and comprehensive treatment to give supplement to fault characteristic parameters preliminary,and gives the method for extracting the fault classification rules and process of this model.This method realizes the comprehensive utilization of fault information,and fuses the rough set that has good performances in incomplete data and complex patterns depicting and support vector machine that has excellent generalization performance effectively.The fault diagnosis example shows that the method can effectively improve the efficiency of fault diagnosis.A fault diagnosis model based on multi dimension characteristic parameters is constructed.The electrical characteristics and oil test data are set as the characteristic parameter in fault diagnosis to solve the problems that fault information is lack and fault information carried is limited,and the characteristic evaluation and kernel principal component analysis method are used to reduce the dimension of characteristic fusion to realize the complementary and fusion of characteristic information.The example analysis shows that this method can not only solve the problem of single parameter,but also has a good effect in diagnosing.In multidimensional characteristic information fault diagnosis model,for the limited ability of single intelligent algorithm in fault diagnosis,based on the concept of integration in information fusion,this paper considers using D-S evidence theory to get over limitation and build a multi-level information fusion fault diagnosis model based on multi dimension characteristic parameter and fault decision in order to solve the problem that information is single and method is single meanwhile.An example is given to demonstrate the accuracy and feasibility of the proposed method.