Zhiqiang Yang: Power Transformer Fault Diagnosis Based on Mpso-svm

This paper builds the power transformer diagnosis model based on the improved particle swarm optimization— support vector machine (MPSO-SVM). Speed update and particle self-adaptation self-variation are introduced to optimize the standard particle swarm algorithm, thus overcoming the defect of the standard particle swarm optimization algorithm and increasing the power transformer fault diagnosis accuracy rate of SVM. Through the analysis of the relationship between the transformer fault and the dissolved gas, the volume content of the dissolved gas of the transformer is adopted as the fault feature index. Through the experiment numerical analysis, results suggest that: the test sample recognition accuracy of the model parameters acquired by MPSO-SVM is higher than that acquired by the standard PSO by 17.86%.