In order to realize fault prediction of equipments,the paper uses the Support Vector Machine (SVM) as the basic learning algorithm,and adopts the weighted SVM regression method to assign relatively larger weight to the mutation point,so as to enhance the training of the mutation point and improve the fault prediction accuracy.By using the adaptive weight clipping method,the sample points with smaller weights are eliminated by calculating the regression weights of the sample points.Meanwhile,the number of samples participating in the training is reduced so as to improve the forecasting speed.Finally,the fault prediction model is established by selecting the appropriate kernel function and related parameters.The validity and superiority of the algorithm are verified by taking a certain type of communication station as an example.