Research on Power Equipment Defect Prediction Method Based on RS-SVM

Aiming at the problems of power equipment defect prediction in power grid, a method based on rough sets and support vector machine (RS-SVM) is proposed for defect prediction. Rough sets (RS) work as a preprocessor in order to obtain the minimum condition attribute set before the sample data are processed by support vector machine (SVM), and the idea of classification is used to predict the power equipment defect. Based on the equipment defect information data provided by a grid company, the dimensions of the original data sets are reduced from 14 to 6 by RS. The experimental results show that the prediction performance of the proposed method is better than other methods, regardless of whether or not the attribute reduction is carried out. In addition, the performance of the four methods after attribute reduction is improved.