基于格兰杰因果检验和主成分回归分析的月发电量预测  

Monthly Generated Energy Forecast Based on Granger Causality Test And Principal Component Regression Analysis

作　　者：刘航[1] 杜依航[1] 阮亮[1] 吴忠群[1] 

机构地区：[1]华北电力大学经济与管理学院,北京102206

出　　处：《陕西电力》2014年第2期55-60,共6页Shanxi Electric Power

基　　金：国家自然科学基金重点资助项目(70901025)

摘　　要：利用格兰杰因果检验中不同滞后期的自变量对因变量的因果关系的不同,对原始数据进行滞后处理,通过不断试错,选取对发电量影响显著的5个因素作为自变量,结合主成分分析法,提取第一、二主成分与发电量进行回归分析,并对我国发电量进行了预测。通过与不作处理的主成分回归预测结果和向量自回归预测结果进行对比,发现该方法在对月发电量进行预测时具有更高的准确率。By using the difference between causal relationships of the variables to the dependent variable with different lag in Granger causality test,hysteresis processing is conducted on the original data and five factors which have obvious influence is selected as independent variables.Combined with principal component analysis method,the first and second principal component is forecasted and regression analysis is made with generated energy,then generated energy is forecasted.Contrasted the results of none processed principal component regression with vector autoregression forecasting,it is found that the method has higher accuracy rate on monthly generating capacity forecast.

关 键 词：因果检验 主成分分析 负荷预测 

分 类 号：TM615[电气工程—电力系统及自动化] F407.61[经济管理—产业经济]