Short-term Load Forecasting Based on PCA-SVM

According to the influence of input variables on the prediction precision and generalization capability of support vector machine algorithm,principal component analysis,which can eliminate the collinearity of variables,is used to accomplish data preprocessing by extracting characteristic information from training data set.Case studies using the history data offered by East-Slovakia Power Distribution Company show that PCA-SVM algorithm has good performances in reducing the dimension of the input-space as well as increasing forecasting accuracy.