基于数据统计特性考虑误差修正的两阶段光伏功率预测  

Two-stage Photovoltaic Power Forecasting and Error Correction Method Based on Statistical Characteristics of Data

作　　者：刘杰[1] 陈雪梅 陆超[1] 毛航银 LIU Jie;CHEN Xuemei;LU Chao;MAO Hangyin(Department of Electrical Engineering,Tsinghua University,Haidian District,Beijing 100084,China;State Grid Zhejiang Electric Power Research Institute,Hangzhou 310014,Zhejiang Province,China)

机构地区：[1]清华大学电机工程与应用电子技术系,北京市海淀区100084 [2]国网浙江省电力有限公司电力科学研究院,浙江省杭州市310014

出　　处：《电网技术》2020年第8期2891-2897,共7页Power System Technology

基　　金：国家电网公司总部科技项目(52110419003R)。

摘　　要：提高光伏功率预测的准确性有助于电网调度计划的制定,对于电力系统安全稳定和经济运行具有重要意义。基于数据的统计特性分析,提出一种考虑误差修正的两阶段光伏功率预测模型。首先,利用主成分分析(principal component analysis,PCA)克服因气象因素间相关性导致的回归模型共线性问题,通过波动量分析和聚类分析建立了输出功率与气象类型之间更精细的匹配模型;然后采用回归分析的方法构建了各子集对应的多重线性回归模型,实现光伏功率的初步预测;最后根据初步预测误差的分布特性,建立了更为准确的初步预测误差概率分布模型,实现初步预测结果的误差修正。基于实际光伏功率曲线和气象数据的算例结果验证了所提方法的有效性。Accurate photovoltaic power(PV) prediction contributes to the efficient power grid dispatching and the reliable operation. In this paper, a two-stage PV prediction model with error corrections based on the statistical analysis is presented. Principal component analysis(PCA) is adopted to remove the collinearity of various meteorological values. Through fluctuation and cluster analysis, a novel model is proposed with more precise fittings to the output power and the meteorological types. Then, a preliminary prediction on PV power can be obtained by multiple linear regression models corresponding to each subset. To correct the errors further, the distribution characteristics of these errors are also analyzed. The validations based on practical PV power and meteorological data show the advantages of the proposed forecasting method.

关 键 词：光伏功率预测 主成分分析 聚类分析 多重线性回归 误差修正 

分 类 号：TM721[电气工程—电力系统及自动化]