摘要
提出了一种基于最小二乘支持向量机的风电场风速预测方法。以历史风速数据、气压、温度作为输入,对风速和环境条件进行训练,建立预测模型,并且运用网格搜索法确定模型参数。算例结果表明,使用上述方法预测的风速与真实值基本一致。将本文提出方法与BP(back propagation)神经网络法的预测结果进行对比,表明前者具有更高的精度和更强的鲁棒性,因此是一种比较有价值的风速预测方法。
A wind speed forecasting for wind farm based on least squares support vector machine （LS-SVM） is proposed. Taking historical wind speed data, atmospheric pressure and temperature as the input, the wind speed and environmental condition are trained by LS-SVM, then a forecasting model is built, and by use of grid search the parameters of the model are determined. Forecasting wind speed by the proposed method, the obtained results are basically in accordance with the values of actual wind speed. Comparing the wind speeds forecasted by the proposed method with those forecasted by BP neural network based method, it is shown that the proposed method is better than the latter in robustness and forecasting accuracy.
出处
《电网技术》
EI
CSCD
北大核心
2008年第15期62-66,共5页
Power System Technology
基金
国家重点基础研究发展计划项目(973项目)(2004CB217908)。~~
