基于AP-SVM多模型建模的风电场负荷预测研究

Power Prediction Research of Wind Farm Based on AP-SVM Multi-model Modeling

针对风电场中期负荷预测模型复杂多工况的情况,提出了基于仿射传播聚类和最小二乘支持向量机的多模型建模负荷预测方法。该方法先用仿射传播聚类算法对样本聚类,再用最小二乘支持向量机算法进行子模型建模。测试样本先根据相似性的度量方法进行归类,再用其所属子模型进行预测输出。最后利用某风场数据进行了建模和预测实验,结果表明该多模型建模方法有较高的预测精度和良好的泛化能力。

Characteristics of multiple conditions are existed in the medium-term power forecast model of wind farm. For this problem, a multi-model modeling method based on affinity propagation clustering and LS-SVM algorithm is presented. In this method,affinity propagation clustering algorithm is used to cluster the training sample. Then, the sub-models are trained by LS-SVM. The predicted values of the testing sam- ples are forecasted by the sub-models after they are classified by the similarity measurement. Finally,mod- eling and prediction experiment is arranged by using the field data. The experiment shows that the proposed method has high prediction accuracy and good generalization ability.