基于双层随机森林算法的短期负荷预测模型  

Short-term Load Forecasting Model Based on Double-layer Random Forest Algorithm

作　　者：邢书豪 高广玲 张智晟 XING Shuhao;GAO Guangling;ZHANG Zhisheng(College of Electrical Engineering,Qingdao University,Qingdao,Shandong 266071,China;State Grid of China Technology College,Jinan,Shandong 250000,China)

机构地区：[1]青岛大学电气工程学院,山东青岛266071 [2]国网技术学院,山东济南250000

出　　处：《广东电力》2019年第9期160-166,共7页Guangdong Electric Power

基　　金：山东省电力科技计划项目(2019);2016智慧青岛建设计划重点项目(强化重点领域智慧企业服务类-11)

摘　　要：为了提高电力系统短期负荷预测的准确性,提出了基于双层随机森林算法的电力系统短期负荷预测模型。在预测模型的第1阶段,用随机森林算法构建初始预测模型,得到训练残差,将训练残差代入原始训练样本构建新的训练样本;在第2阶段,采用新的训练样本对随机森林算法再次进行训练,从而充分解读训练样本中的有效信息;最后,将2个阶段的随机森林模型融合,得到双层随机森林预测模型。以某市实际电力负荷数据作为算例,对模型进行仿真验证,结果表明:相比于基于单层随机森林算法的预测模型,基于双层随机森林算法的预测模型准确性更高。To improve accuracy of short-term load forecasting of the power system,a short-term load forecasting model for the power system based on the double-layer random forest algorithm is proposed.In the first stage of the model,the random forest algorithm is used to construct the initial forecasting model and obtain training residuals which are substituted into the original training sample to construct a new training sample.In the second stage,the new training sample is used for retraining the random forest algorithm so as to fully understand effective information of the training sample.Finally,the random forest models in two stages are integrated and a double-layer random forest forecasting model is obtained.Actual power load data of one city is taken as an example for simulation verification and the result indicates that compared to the forecasting model based on the single-layer random forest algorithm,the forecasting model based on the double-layer random forest algorithm has higher accuracy.

关 键 词：随机森林算法 电力系统 短期负荷预测 Bagging方法 CART决策树 

分 类 号：TM715.1[电气工程—电力系统及自动化]