基于偏最小二乘回归分析的中长期电力负荷预测  

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

作　　者：毛李帆[1] 江岳春[1] 龙瑞华[1] 李妮[1] 黄慧[1] 黄珊[1] 

机构地区：[1]湖南大学电气与信息工程学院,湖南省长沙市410082

出　　处：《电网技术》2008年第19期71-77,共7页Power System Technology

摘　　要：针对中长期电力负荷预测,介绍偏最小二乘回归分析方法的原理,推导该算法的简化建模步骤。该方法能在最大限度保留原有数据信息的前提下,将数据信息集中在几个互不相关的主成分上,因而能有效解决建立负荷预测模型时由于样本个数较少及自变量存在严重的多重相关性,难以通过多元回归分析建立预测模型的问题。通过算例对偏最小二乘回归分析方法、最小二乘法和逐步回归分析方法进行了比较,结果表明,将偏最小二乘回归分析方法用于中长期电力负荷预测时,计算快捷,准确性高,具有较强的实用性。Aiming at medium- and long-term load forecasting, the principle of partial least squares regression analysis is presented and the simplified modeling procedures of this algorithm is derived in detail. Under the presupposition of keeping original data information as possible, the proposed method concentrates the data information to several principal components uncorrelated each other, thus the difficulty of not enough sample numbers and severe multiple correlation of self-variables that makes it hard to build forecasting model by multiple regression analysis can be effectively solved while the load forecasting model is built. By means of calculation example, the partial least squares regression analysis method, the least squares method and the stepwise regression analysis method are compared, and comparison results show that as far as medium- and long-term load forecasting, the proposed modeling method is fast, accurate and practicable.

关 键 词：中长期负荷预测 偏最小二乘回归分析 成分提取 多元线性回归 

分 类 号：TM715[电气工程—电力系统及自动化]