Research On Power Transformer Fault Diagnosis Model And Algorithm Based On Dissolved Gas Analysis

In the power system,power transformer plays an important role.If transformer has fault,it will cause instability of power supply,serious economic losses,and even endanger people’s life safety.Therefore,it is necessary to study a good method for fault diagnosis of power transformer to ensure the safe and stable operation of power system.The fault diagnosis of oil immersed transformer is mainly based on the data of dissolved gas analysis（DGA）.With the development of artificial intelligence technology,fault diagnosis technology based on support vector machine（SVM）has been concerned by a large number of researchers,but it also has some disadvantages,such as fuzzy DGA data boundary and difficult to determine hyperparameters of SVM.Aiming at the above shortcomings,this paper proposes a power transformer fault diagnosis method based on kernel principal components analysis（KPCA）and hybrid improved seagull optimization algorithm（SOA）optimize SVM（TISOA-SVM）.Firstly,this method considers using KPCA which has a good feature extraction effect on nonlinear data to process DGA feature quantity,and then further proposes TISOA optimize SVM.The improvement methods for SOA are as follows: 1)The improved tent mapping（Mtent）is used to replace the original population initialization method to improve the diversity of the initial population of seagulls;2)Nonlinear inertia weight is proposed to improve the optimization accuracy and efficiency of SOA;3)The original foraging formula of SOA is studied,and the random double helix formula is innovatively proposed to further improve the optimization accuracy and convergence speed of SOA.At the same time,in order to verify the optimization performance of TISOA,23 traditional benchmark test functions and CEC2015 benchmark test functions are used to test TISOA and six meta-heuristic algorithms to verify TISOA optimization performance.The results show that TISOA has excellent optimization performance.Finally,the fault diagnosis method based on KPCA and TISOA-SVM is obtained,and Three examples are used to verify the performance of the proposed methods.The results show that the proposed method has higher diagnosis accuracy and efficiency,and has stronger validity and significance.