Analysis Of Transformer Operation State Based On SVM

Transformer is an important part of the Grid,and its running state and operation condition are directly related to the stability of the Grid.The analysis of transformer operation state which can be used to predict the failure and ensure the normal operation of the Grid has great significance.These running data provided by State Grid ** power supply company which includes voltage,current,active power and reactive power,etc.is closely related to the state of the main transformer.First,we used principal component analysis to reduce dimension,and the simulation results show that the performance of Running state analysis by the main component analysis is better.We selected support vector machine to realize the transformer operation status analysis based on the characteristics of the data and used the linear kernel function,the polynomial kernel function and the radial basis kernel function respectively to analyze the data,and then calculated the parameter optimization of the above-mentioned kernel functions with particle swarm algorithm,grid algorithm and genetic algorithm,three kinds of optimization methods were proposed.The result of the experiment showed that the model had the highest accuracy and the shortest time with the radial basis kernel function support vector machine.Finally,we used the logistic regression and neural network to perform the operation state analysis to the same data set,and the experimental result showed that the support vector machine model has the best analysis effect,which fully demonstrates that the support vector machine had certain advantages and better effects in the operation state of the main transformer.We can also introduce some new transformer state characteristics(temperature,noise,etc.)and the new algorithm to analysis of transformer running state more in-depth research which ensure continuous and stable operation of the whole power system in subsequent work.