摘要
通过对中国典型城市居民1992~2006年生活用电及其影响因素数据的聚类分析，得出中国城市居民生活用电在不同收入水平下的用电特征，首先通过分析各个阶段居民拥有电器种类及各种电器的增长率，对不同阶段的电器赋予权重，得到一个综合电器指数作为预测的输入端。其次，基于面板数据模型对不同类型用电居民的生活用电做预测，并得出预测误差，不仅考虑了数据的时间效应还考虑了截面效应，也避免了由于数据序列不足而造成的用电特点不明显。通过对中国典型城市的实证分析，论证了该方法的有效性。
The electricity consumption characteristics of urban residents under different living conditions were identified by cluster analysis of the data on residential electricity consumption and relevant influencing factors of typical cities in China from 1992 to 2006. Based on the analysis of the main kinds of household appliances used under different living conditions as well as the growth rate of these appliances, weights were endowed to these appliances and the comprehensive index of appliances was got. Moreover, the Panel Data model was used to forecast the residential electricity consumption, and the forecast errors were also given. With the methods, not only the time effect of data but also the section effect were considered, and the problem caused by the lack or too short of data series could also be avoided. Demonstration analysis of Chinese typical cities proved the method effective.
出处
《华东电力》
北大核心
2009年第1期51-55,共5页
East China Electric Power
基金
国家自然科学基金项目(7067104270771039)
教育部人文社科基金(07JA790092)
关键词
PANELDATA
城市
居民用电
预测
变系数模型
Panel Data
city
residential electricity consumption
forecast
varying-coefficient model
