Design And Research Of Transformer Fault Diagnosis Method

Transformer is one of the key components that affect the operation of the power system.The working condition of the transformer directly determines whether the power system is running well or not.If there is a problem with the operation of the transformer,it will seriously affect the power supply and distribution of the power system.Here,in order to better study the problems of transformer operation,the author proposes corresponding solutions based on this,and puts forward the transformer fault diagnosis algorithm based on firefly algorithm to optimize BP neural network.Diagnostic research.The main research work of this paper is as follows:The paper analyzes the generation mechanism of gas in transformer oil and the correspondence between characteristic gas and fault type in transformer oil.Therefore,BP neural network algorithm is proposed for fault diagnosis of power transformer.Aiming at the problem that BP neural network is easy to fall into local minimum point and slow convergence rate,it is based on the firefly optimization BP neural network algorithm.The GSO-BP algorithm improves the overall search ability of the BP algorithm and improves the optimization efficiency of the algorithm.GSO-BP can greatly reduce the local optimal defects of the latter algorithm and improve its diagnostic quality and efficiency.Since the load capacity of the power transformer and its usable useful life depend on its thermal characteristics,after systematically analyzing the internal heating process and temperature distribution law of’the power transformer,combined with the specific conditions of the transformer’s various load conditions and cooling methods,The simplified transformer internal temperature calculation prediction model is used.Based on this,the MP-ELM algorithm is used to establish the transformer surface temperature prediction model,which provides theoretical basis and judgment basis for state maintenance.Finally,the simulation verification of the algorithm proves that the GSO-BP algorithm can effectively improve the speed and accuracy of transformer fault diagnosis.The MP-ELM model has been continuously tested in practice and can be better used in fault detection.Moreover,compared with the ELM neural network,the model has the advantages of strong stability and high prediction accuracy.Through experimental verification,the proposed method in transformer fault monitoring and diagnosis method can effectively improve the practicability of transformer fault diagnosis system,and provide a new reference for the improvement and improvement of transformer fault diagnosis.