摘要
在能源安全、环境污染和气候变化三大国际问题下,如何提高多能负荷预测精度成为全球的焦点问题。为此,提出了一种新颖的基于梯度提升回归树(GBRT)和轻型梯度提升机(LGBM)的多能负荷组合预测方法。首先,利用Spearman相关系数分析多能负荷及相关因素间的相关性,构建更有效的模型输入数据集;然后,利用网格搜索及交叉验证方法,分别对GBRT模型和LGBM模型进行训练,得到针对不同类型负荷的最优参数组合的模型;最后,将两种模型进行加权组合,得到最终的多能负荷预测结果。算例分析结果表明,所提模型能结合两种模型的优势,深入挖掘不同负荷之间的耦合关系,与其他模型相比有更高的多能负荷预测精度和更好的预测适用性。
Under such three major international problems as energy security,environmental pollution and climate change,how to improve the accuracy of multi⁃energy load forecasting has become a global focus.To this end,a kind of novel multi⁃energy load combined forecasting method based on gradient boosting regression tree(GBRT)and light gradient boosting machine(LGBM)is proposed.Firstly,the Spearman correlation coefficient is used to analyze the correlation between multi⁃energy loads and related factors so to construct a more effective model input data set.Secondly,the the grid searching and cross validation method is used to train the GBRT and LGBM model respectively to obtain the model combined with optimal parameters for different types of loads.Finally,the two models are weighted and combined to obtain the final forecasting results of multi⁃energy loads.The calculation and analysis results show that the proposed model can combine the advantages of GBRT model and LGBM model,can deeply explore the coupling relations among different loads.Compared with other models,the proposed model has higher accuracy and better applicability of multi⁃energy load forecasting.
作者
刘晓东
常飞
王璇
许若冰
LIU Xiaodong;CHANG Fei;WANG Xuan;XU Ruobing(State Grid Jiangsu Electric Power Co.,Ltd.,Nanjing Power Supply Corporation,Nanjing 210019,China;Tianjin Xianghe Electric Co.,Ltd.,Tianjin 300000,China)
出处
《电力电容器与无功补偿》
2023年第3期97-102,共6页
Power Capacitor & Reactive Power Compensation
基金
国网江苏省电力有限公司科技项目(J2019001)。
关键词
梯度提升决策树
轻型梯度提升机
网格搜索
交叉验证
多能负荷预测
gradientboostedregressiontrees
lightgradientboostingmachine
gridsearch
cross⁃validation
multi⁃energy load forecasting
