摘要
进行实际短期负荷预测时 ,对某个固定地区 ,用不同预测方法可能得到不同的预测结果。文中基于对这些不同的预测结果的分析 ,提出了以得到一个唯一的综合预测曲线为目标的优化模型。根据模式识别的基本原理 ,这个新模型以待预测日与历史日之间相关因素的最佳匹配为原则。通过虚拟预测结果与实际负荷曲线数据的误差平方和的最小化 ,可以得到综合模型中各种单一预测结果所占的权重值大小。实际算例研究表明 。
For a given electric power utility. various forecasting results can be obtained by using different models. Based onthe analysis of these various results. the paper presents an optimization model so that a unique integrated forecasting loadcurve can be got. According to the basic principles of pattern recognition. the new model concentrates on the optimal matchbetween the different correlated factors of future and past days. By minimizing the weighted matching error of the hourlyload, the optimal weight value is identified. Comparison of forecasting results obtained by the integrated model and linearregression method shows the efficiency of the proposed method.
出处
《电力系统自动化》
EI
CSCD
北大核心
2000年第9期42-44,共3页
Automation of Electric Power Systems
关键词
短期负荷预测
模式识别
综合模型
电力系统
short term load forecasting
pattern recognition
cluster analysis
integrated model
