Transformer Based On Artificial Bee Colony-support Vector Machine Optimization Algorithm Troubleshooting

As the core component of the whole system,the power transformer is inseparable from the operation of the entire power grid.Once the power transformer fails,it will bring great threat to the people and the country's property.Therefore,the research of power transformer is the most important.weight.Because there are fewer faults in the transformer,resulting in a small number of fault data samples,that is a small sample data.For this problem,it is found that the support vector machine has superiority in dealing with multi-classification problems of small samples.First,by using the dissolved gas analysis method(DGA technology),the fault data is collected,and the absolute percentage of the dissolved gases in the oil which has a large influence on the transformer fault is selected,and selected according to the characteristics of the data and the characteristics of the algorithm.The optimization method and classifier type are used to construct a diagnostic model of improved bee colony algorithm to optimize the parameters of support vector machine-IABC-SVM.In the intelligent fault diagnosis method,the input characteristic quantity is usually the percentage of the dissolved gas,and it is often ignored whether the pre-processing of the data affects the correct rate of the fault diagnosis,and the fault diagnosis correct rate is obtained by comparing whether the pre-processing is performed.The obtained diagnostic accuracy rate is indeed greatly affected by the data pre-processing.After the original fault sample data is processed by the data transformation and the feature selection method,the fault feature quantity required for the diagnosis is determined.Secondly,in order to better solve the shortcomings of the artificial bee colony algorithm,for example,there will be local extremum problems,the search efficiency is relatively low,etc.For these problems,two improved methods are proposed for the artificial bee colony algorithm.The two-dimensional uniform method is used to initialize the bee colony,which can evenly distribute the food source within the range of all solutions,which not only increases the diversity of the artificial bee colony,but also makes the algorithm achieve better search rate and global search ability;It is an update strategy to improve the food source.The Euclidean distance method is used to enable the food source to automatically update the food source,thereby improving the convergence of the algorithm.Finally,through the comparative analysis of the fault diagnosis accuracy obtained by using three different fault diagnosis models of GA-SVM,PSO-SVM and IABC-SVM,in order tomake the diagnosis result more convincing,this chapter uses the test sample data to conduct multiple tests.Diagnostics,the result show that IABC-SVM has a more accurate recognition effect for transformer faults.The paper has 13 pictures,27 tables,and 71 references.