Abstract:In order to solve the shortcoming of the support vector machine (SVM)'s penalty factor c and the nuclear function parameter g only can depend upon the apriori knowledge,a model of SVM parameter optimization was presented,based on the optimal algorithm of PSO.Through automatic optimization of the PSO,the optimal parameters of SVM were gotten,and the fault diagnosis was carried on by using the JZQ250 gear box.Experiments show that model presented is a very good solution to the problem of selecting parameters,and makes some promotion of the SVM performance.