Research On Transformer Fault Diagnosis Model Based On AdaBoost And IGWO Optimized SVM

Power transformer is an important hub of power distribution and transmission in the power grid,with the rapid investment and construction of the national integrated energy grid,power transformers move towards higher voltage levels and capacities in response to the needs of the times,if a transformer fails,it may cause serious damage to the resources and economy of the grid,and involve people’s safe living with electricity.(1)In order to improve the accuracy of transformer fault diagnosis by support vector machine(SVM)model,an IGWO-SVM fault diagnosis model is established.An improved grey Wolf algorithm is proposed to optimize the two parameters C and g of SVM.A quasi-inverse learning strategy is used to increase the diversity of the population,helping to improve the solution accuracy and convergence speed of the algorithm;a linear convergence factor is reconstructed to coordinate the balance between global search and local exploitation;a multi-elite search strategy is introduced to avoid falling into local optima and speed up the convergence speed.Then nine test functions are used to verify its good optimization performance.Finally,the transformer fault diagnosis model of IGWO-SVM is constructed.(2)Aiming at the problem that the fault diagnosis model of a single IGWO-SVM transformer has low accuracy and weak stability,the Ada Boost algorithm is used to integrate the IGWO-SVM,and by calculating the error rate of different samples,changing the weights,obtaining a new sample distribution,and then iteratively training,and then obtaining a strong classifier,the transformer fault diagnosis model based on Ada Boost-IGWO-SVM is constructed,and determine the number of weak classifiers for the Ada Boost integration algorithm.(3)Using the non-coding ratio method,the characteristic gases and their proportion are reconstructed into 13 parameters as the input characteristics of Ada Boost-IGWO-SVM transformer fault diagnosis model for simulation and analysis,and the accuracy of Ada Boost-IGWO-SVM model in transformer fault diagnosis is 93.33%,compared with IGWO-SVM,Ada Boost-SVM,PSO-SVM and GWO-SVM,the transformer fault diagnosis model is improved by 3.33%,10%,11.66% and 6.66%,respectively.It proves that the Ada Boost-IGWO-SVM transformer fault diagnosis model can effectively improve the fault identification rate.In addition,the paper also analyzes the stability and comparative analysis of running time of the above model,verifies that the average diagnostic accuracy of the Ada Boost-IGWO-SVM transformer fault diagnosis model proposed in the paper is 92.83% and the standard deviation is0.95,which is better than other transformer fault diagnosis models in terms of mean and standard deviation performance,indicating that the stability of its model is strong.The thesis has 32 pictures,18 tables and 85 references.