Transformer Fault Diagnosis Method Based On Improved Support Vector Machine And Kriging Model

Transformer is the most critical hub equipment in the power system.Once the fault occurs,it will not only affect the reliability of power supply,but also pose a serious threat to the safe and stable operation of the power system,which will lead to serious economic losses.Therefore,the diagnosis of transformer faults can detect potential transformer problems in advance,reduce the economic losses caused by transformer faults,and make the power system run more reliably and safely.In this paper,the corresponding relationship between transformer fault types and characteristic gases is studied,and the dissolved gas analysis method is adopted.On this basis,two different diagnosis methods are proposed,namely,the improved support vector machine based on cuckoo algorithm and the improved kriging model transformer fault diagnosis method based on cuckoo algorithm.The main research contents of this paper are as follows:(1)Collect the sample data,test data,the least squares support vector machine classification model and modeling steps is given,using the cuckoo algorithm for least squares support vector machine nuclear parameter C g and punish parameters optimization,in order to improve the optimization ability of the cuckoo algorithm,inertia weight to improve the cuckoo algorithm,and with the steepest descent method,a new algorithm for parameter optimization.The model is trained and simulated on MATLAB software platform.By comparing genetic algorithm and cuckoo algorithm,the superiority of this algorithm is verified.(2)Is put forward based on the kriging model of transformer fault diagnosis,according to the collected data to create kriging model of faultdiagnosis,the cuckoo algorithm to improve the optimization of related parameters in the model,the use of MATLAB for training model,through examples of simulation show that the computational cost and the fault diagnosis model,the advantage of high fitting precision,and by comparing the genetic algorithm,the cuckoo algorithm of kriging model related parameters optimization method,to verify the effectiveness of the proposed method.