基于偏最小二乘回归的中长期电力负荷预测  

Medium and Long Term Load Forecasting Based on Partial Least-Squares Regression

作　　者：蒋惠凤[1] 何有世[2] 杨伟雄[2] 

机构地区：[1]常州工学院经济与管理学院,常州213022 [2]江苏大学工商管理学院,镇江212013

出　　处：《电力系统及其自动化学报》2007年第5期110-113,共4页Proceedings of the CSU-EPSA

摘　　要：为了支持将来的经济发展和不断满足电力需求,负荷预测已成为电力部门的重要任务,而提高预测精度是负荷预测的关键问题。为此,判断了影响负荷的经济因素之间存在的多重共线性,用偏最小二乘回归方法消除其共线性影响,并建立了预测模型。结果表明,该方法能准确地估计出变量的回归系数,能避免使用普通最小二乘回归时出现的异常回归系数,预测的相对误差平均为9.83%,最小相对误差为-0.01%。To support economic growth and meet power requirement continually in the future, load forecasting has become a very important task for electric utilities, and the forcasting precision is the key problem. The multiple-colinearity among economic factors influencing load is judged and the effect of multiple-colinearity is weakened through partial least-square regression. Then a forecasting model is established. Results show that the proposed method can well estimate the variable＇s regression coefficients and avoid the abnormal regression coefficionts while using conventional least-square regression. The mean relative error of forcastng is 9. 83%, and the smallest is -0. 01%.

关 键 词：负荷预测 多重共线性 方差膨胀因子 偏最小二乘回归 

分 类 号：TM7[电气工程—电力系统及自动化]