| The power transformer is responsible for the power transmission inside the power grid,as well as the transmission of electric energy from the power grid to the terminal distribution network,the voltage rise and fall and energy conversion of systems at all levels,etc.,and it plays the role of a hub in the entire power grid.It is an essential part of the power system.The safety of electrical equipment is directly related to the reliability of the power system.However,many transformers are currently facing serious aging problems,so the probability of failure is gradually increasing.If a large oil-immersed transformer fails,it will cause damage to the equipment,and cause largescale power outages or even casualties.Therefore,the fault diagnosis of oil-immersed transformer is indispensable.Based on the existing research of many scholars,this paper has made the following innovative achievements in the fault diagnosis and fault warning of oilimmersed transformers based on unbalanced data:a transformer fault diagnosis method based on ADASYN-AdaBoostSVM algorithm is proposed,The accuracy of transformer fault diagnosis is improved,and a transformer gas content prediction method based on ARIMA algorithm is proposed.Based on the existing research,this paper summarizes the fault characteristic variables that can reflect the state information of the transformer,and uses the support vector machine algorithm to diagnose the fault of the transformer,which can diagnose the fault state of the transformer in time.In addition,this paper proposes a support vector machine transformer fault diagnosis method based on AdaBoost algorithm enhancement,and the accuracy of diagnosis is further improved through the optimization of the integrated algorithm.The support vector machine algorithm enhanced by the AdaBoost algorithm can quickly judge whether the transformer is faulty,and provide a reference for subsequent maintenance strategies;in the study,it was found that although the SVM algorithm optimized by the integrated algorithm has a higher overall accuracy of transformer fault diagnosis However,the diagnostic accuracy of each fault type is quite different.In order to accurately identify the specific faults generated by the transformer,a transformer fault diagnosis method based on the ADASYN oversampling algorithm is proposed based on the unbalanced data of the transformer.,by oversampling the fault data of the minority category,the accuracy of the algorithm in the classification of imbalanced data is improved.Finally,the ARIMA algorithm is used to predict the contents of the five characteristic gases of the transformer,and the optimal parameters of each gas prediction model are selected through the Akaike Information Criterion(AIC).After verification,the average error and relative error of the predicted results are small,it shows that the prediction accuracy of the model for gas content in oil is high,and the predicted gas value is substituted into the established AS-AdaBoostSVM transformer fault diagnosis system,which can give early warning of possible faults in the future,so as to provide Make a maintenance plan afterward. |
