摘要
为实现准确的风电功率预测,提出了一种基于提升回归树与随机森林的风电功率集成预测方法。在建立预测模型的第1阶段,首先使用提升回归树建立初始的预测模型,然后计算训练残差;在第2阶段,采用随机森林算法,对训练集上得到的残差进行学习。两阶段的模型相加,得到最终的预测模型。以IEEE能源预测工作组提供的风电功率预测数据进行了测试,所提预测模型得到的结果的均方根误差为0.148 8,相比其他方法,预测精度有明显的改善,结果表明,该预测方法可以有效地提取数据中包含的信息。
To improve the forecasting accuracy of wind power generation,a wind power generation ensemble forecasting method based on boosted regression tree and random forest is proposed in this paper.On the first stage of building the forecasting model,boosted regression tree is used to build an initial forecasting model at first.Then,the training residuals are computed.On the second stage,random forest method is used to learn residuals obtained from the training set.At last,the models on the two stages are combined to build the final forecasting model.The forecasting data of wind power generation provided by IEEE working group on energy forecasting are tested.The root mean squared error of the proposed forecasting model is 0.148 8.Compared with the other models,the forecasting accuracy is improved obviously,showing that the proposed method can extract the useful information contained in the data efficiently.
作者
李国
江晓东
LI Guo;JIANG Xiaodong(School of Electrical and Information Engineering,Tianjin University,Tianjin 300072,China;School of Electrical and Computer Engineering,Cornell University,Ithaca 14853,USA)
出处
《电力系统及其自动化学报》
CSCD
北大核心
2018年第11期70-74,共5页
Proceedings of the CSU-EPSA
基金
国家自然科学基金重点资助项目(51337007)
关键词
风力发电
集成预测
提升回归树
随机森林
wind power generation
ensemble forecasting
boosted regression tree
random forest
作者简介
李国(1984—),男,博士研究生,研究方向为数据分析、可再生能源预测及其在电力系统中的应用。Email:li_guo@tju.edu.cn;江晓东(1956-),男,博士,教授,博士生导师,研究方向为电力系统稳定和控制、非线性系统理论及其应用、全局优化技术和应用。Email:chiang@ece.cornell.edu。
