Power Transformer Fault Diagnosis Using Support Vector Machine and Particle Swarm Optimization

Dissolved gas analysis is a simple and effective methods for incipient fault detection in power transformers. To enhance fault diagnosis capability of related interpretation approaches, a new fault diagnosis model is proposed in this paper. Feature selection techniques are used to select the most informative features. Particle Swarm Optimization algorithm is integrated with support vector machine to optimize key parameters. Fault diagnosis model is established and tested based on the best features and the optimal parameters. The result and comparison show that the accuracy of PSO-SVM method has higher accuracy than that of conventional methods.