Research On Transformer Fault Diagnosis Method Based On VSRP And Grey Wolf Optimized SVM

With the development of society,the higher the requirement for the safety and reliability of power supply.The security and stability of the power system is closely related to the operation state of the transformer.Once the transformer fails,it will have a significant impact on the power system.Therefore,for the core equipment transformer fault diagnosis,it has more and more important research significance.Firstly,to solve the problem that the high-dimensional and complex original fault data of transformer affect the diagnosis accuracy,very sparse random projection(VSRP)was used to reduce the dimension.In order to verify the effectiveness of the proposed method,the separability parameters and visual classification effect of VSRP were compared with principal component analysis(PCA),linear discriminant analysis(LDA)and random projection method.The results showed that VSRP was superior to the other three methods in reducing dimension while maintaining data characteristics,the accuracy of diagnosis was also improved significantly.Secondly,aiming at the problem of low accuracy of support vector machine(SVM)for transformer fault classification,β-GWO-SVM fault diagnosis model was established.firstly,the kernel space theory was introduced to implement the nonlinear transformation of SVM to adapt to the nonlinear and inseparable transformer fault;reintroduce β-Chaos sequence was used to optimize the control parameter α in the traditional grey Wolf algorithm(GWO),proposed β-GWO optimization algorithm;finally,we use the influence of β-GWO on two parameters C and γ of SVM.The optimal parameters were applied to transformer fault classification.The verification showed that,β-GWO-SVM model not only overcomed the shortcoming that GWO algorithm was easy to fall into local optimum,but also improved the accuracy and efficiency of SVM in transformer fault diagnosis.Finally,the optimization method and diagnosis model were simulated and analyzed.First,two test function pairs were used to verify the effectiveness of β-GWO optimization algorithm.At the same time,compared with GWO and particle swarm optimization(PSO),the results showed that in terms of convergence speed and iteration times,β-GWO algorithm was the best;then fault data processed by different methods were input respectively β-GWO-SVM model,making fault diagnosis for transformer,the diagnosis results were compared with the diagnosis results of PSOSVM and GWO-SVM.The results showed that the method of paper can diagnose transformer fault information more accurately,quickly,accurately,reliably,and has strong generalization ability.The thesis has 40 pictures,14 tables and 92 references.