POWER GRID FAULT PREDICTION METHOD BASED ON FEATURE SELECTION AND CLASSIFICATION ALGORITHM

Abstract

In order to accurately predict the risk level of power grid with considering weather factors, a fault level prediction method based on Random Forest and Multi-classification Support Vector Machine is proposed. After data preprocessing and fault classification, the relevant meteorological features of distribution network fault are summarized. The features' weights can be calculated and sorted by using Random Forest algorithm, to obtain the optimal fault feature set, and then the fault level is predicted based on multi-classification SVM. Through the analysis of practical examples, the performance of the proposed method is better than that of SVM alone, and the prediction results can provide an effective basis for risk pre control of power grid, which has practical application value