关键词: 最小二乘支持向量机, 模糊核聚类, 集成学习, 短期负荷预测

Abstract: To further enhance the regression prediction accuracy of support vector machine, a Least Squares Support Vector Machine (LS-SVM) ensemble model based on Kernel Fuzzy C-Means clustering (KFCM) was proposed. The KFCM algorithm was used to classify LS-SVMs trained independently by its output on validate samples, the generalization errors of LS-SVMs in each category to the validate set were calculated of the LS-SVM whose error was minimum would be selected as the representative of its category, and then the final prediction was obtained by simple average of the predictions of the component LS-SVM. The experiments in short-term load forecasting show the proposed approach has higher accuracy.

Key words: Least Squares Support Vector Machine （LS-SVM）, Kernel Fuzzy C-Means clustering (KFCM), ensemble learning, short-term load forecasting