基于偏最小二乘回归的山东省电力需求预测分析  

Electricity demand forecast analysis for Shandong province based on partial least square regression

作　　者：于松青[1,2] 林盛[1] 

机构地区：[1]天津大学管理与经济学部,天津300072 [2]国网山东临清市供电公司,临清252600

出　　处：《干旱区资源与环境》2015年第2期14-20,共7页Journal of Arid Land Resources and Environment

基　　金：教育部新世纪人才支持计划(NCET-07-0958)资助

摘　　要：偏最小二乘回归可以对系统信息进行分解和筛选,辨别系统中的信息和噪声,从而能够有效的克服多重共线性在系统建模中产生的不利后果。文中选取GDP、总人口、全社会固定资产投资等10个影响因素,建立偏最小二乘回归模型对山东省电力需求进行预测分析。研究结果表明:电力需求与10个影响因素均具有显著的相关关系,这些影响因素对电力需求的解释能力达到了98.09%。得到的回归方程具有较高的预测精度。其中,第二产业的发展、总人口的增加和GDP的增长为电力需求增加的最主要因素,经济结构的转型和节能减排等政策的实施会一定程度的减缓电力需求增加所带来的压力。最后,对"十二五"规划未来几年山东省的电力需求进行预测。Partial least squares regression can be applied to analyze the information of system,and identify the information and noise of the system. This can overcome the adverse consequences of multicollinearity system modeling. We selected totally 10 influencing factors such as the GDP,total population and total fixed asset investment etc. to establish the partial least squares regression model to forecast the electricity demand in Shandong province. The research results show that the electricity demand and 10 factors were significantly correlated,the explanatory power of the electricity demand reached 98. 09%,and the resulting regression equation had higher prediction accuracy. The development of the secondary industry,the increase in the total population and the growth of GDP were the most important factors of the increasing demand for electricity. The transformation of the economic structure,as well as energy saving will relieve the pressure brought by the increasing electricity demand to a certain extent.

关 键 词：偏最小二乘回归 电力需求 交叉有效性 变量投影重要性 

分 类 号：F407.61[经济管理—产业经济]