Research On Intelligent Fault Diagnosis Of Power Transformer Recognition

As vital equipment in power system, reliable operation of power transformersis essential for the grid security, thereby increasing the transformer fault diagnosisand treatment is of great significance in promoting security and stability operation.There are many fault diagnosis methods, such as artificial neural network, graytheory, fuzzy math, and expert system. The advantages and disadvantages of thesemethods are discussed in this paper and reasonable improvement ideas areproposed.GM (1,1) model and GM (1, m) model of the gray theory method are analysedand improved GM (1, m) model is proposed. It changes the original data sequenceby changing the generating way, so the nature of the sequence is changed in order tomeet the gray model smoothness requirements. Transformer fault diagnosis of BPneural network model is built in this paper. The LM algorithm is used to improvethe neural network and the topology both transformer fault diagnosis based onneural network training samples is put into application. Based on the theory offuzzy mathematics, fuzzy diagnostic model of the transformer obtain diagnosticcharacteristics of the weight vector and fuzzy relationship matrix is established.Completing fuzzy diagnostic module, the Warshall is proposed to ultimatetransformer fault diagnosis fuzzy clustering method.Finally, the study of improved three-ratio method, the neural network method,gray theory and rough set method help design an intelligent expert analysis system,which contains overall design concept of intelligent monitoring and diagnosis,visualization module and expertsanalysis system can greatly improve the accuracyof transformer fault diagnosis.