Power Transformer Fault Diagnosis Model Based On ICA-SVM

In the modern power grid,the safe,economical and stable operation of power transformer plays a vital role in the whole power system and even the whole economic society.When the transformer in operation fails,it often affects the normal power supply in the region and even causes serious economic loss.In order to discover the transformer latent fault in fault occurred and after failure can more accurately determine the types of failure we must be targeted for real-time monitoring of transformer running state prediction in fault diagnosis.Traditional based on dissolved gas analysis(DGA)in transformer oil can be based on the dissolved gas in transformer oil components for early fault diagnosis of transformer.However,as the requirement of modern power system for fault diagnosis accuracy,the DGA analysis method can not meet the requirements,therefore,many experts and scholars have proposed a transformer fault diagnosis method that combines DGA and intelligent diagnostic technology.These methods greatly improve the accuracy of fault diagnosis.This paper illustrate that transformer fault prediction is of great significance,on the basis of discussing the DGA method,three ratio method,Dewey triangle method,and a variety of comprehensive diagnosis method in the field of artificial intelligence applications.The model principle of transformer fault diagnosis based on support vector machine is introduced in detail.After to solve the problem of lack of SVM classification precision,this paper is presented based on the Imperialist Competitive Algorithm(ICA)to optimize the SVM parameters of transformer fault diagnosis model,and set up 118 samples of transformer running state information in advance.By comparing the above sample information with the diagnosis conclusion of the model,the paper verifies that the fault diagnosis model of transformer using ICA algorithm to optimize SVM parameters can successfully improve the diagnostic accuracy.