Early diagnosis of high-voltage circuit breaker failure is an important way of improving the reliability of power system.In this paper

the application of high-performance support vector machine（SVM） to fault diagnosis of high voltage breaker was suggested.For conventional traditional SVM

the selection of parameters in kernel function directly affected diagnosis result.In the present method

for improving the performance of the classification

a novel intelligent particle swarm optimization（PSO）algorithm was adopted to search the optimal parameters.The operating characteristics parameters extracted from the mechanical parameters signals of the high voltage circuit breakers were taken as training and identification samples of support vector machine.The test results show that the present diagnosis method achieves high accuracy of failure diagnosis.