广州市用电量预测模型比较研究  

A Comparative Study on Electricity Consumption Forecasting Models in Guangzhou

作　　者：柯尚军 蔡国田[2,3,4] 张云涛 KE Shangjun;CAI Guotian;ZHANG Yuntao(School of Nano Science and Technology,University of Science and Technology of China,Suzhou Jiangsu 215123,China;Guangzhou Institute of Energy Conversion,Chinese Academy of Sciences,Guangzhou Guangdong 510640,China;Key Laboratory of Renewable Energy,Chinese Academy of Sciences,Guangzhou Guangdong 510640,China;University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学技术大学纳米科学技术学院,江苏苏州215123 [2]中国科学院广州能源研究所,广东广州510640 [3]中国科学院可再生能源重点实验室,广东广州510640 [4]中国科学院大学,北京100049

出　　处：《生态经济》2020年第8期63-67,180,共6页Ecological Economy

基　　金：广东省软科学研究计划项目“广东省能源互联网产业技术路线图”(2016A080803002);广东省科技计划资助项目“资源环境管理辅助决策平台空间大数据分析及应用研究”。

摘　　要：采用灰色模型GM(1,1)等5种方法对2004-2018年广州市用电量作拟合检验并优选应用模型,结果表明:(1)拟合月度用电量时,气温预测法、灰色模型GM(1,1)、随机季节模型和组合预测法的平均相对误差分别为5.44%、3.47%、2.94%和2.45%;拟合年度用电量时,气温预测法、灰色模型GM(1,1)、随机季节模型、人均用电预测法和组合预测法平均相对误差分别为1.87%、2.22%、1.09%、2.36%和1.05%;从拟合精度角度看,灰色模型和随机季节模型较理想,气温预测法和人均用电量法均较差,组合预测法集成各模型优势,效果最佳。(2)运用组合预测法得出2019年、2020年和2021年广州市各月用电量分别平均同比增长3.65%、4.72%和4.65%,年用电量则分别同比增长5.12%、5.18%和5.12%,可用于广州市电网负荷管理。(3)提高预测精度要求更细的用电量和人口统计数据,需要各部门加强数据上报和沟通机制。This paper used six prediction methods such as grey model GM(1,1) to fit the electricity consumption in Guangzhou from 2004 to 2018 and selected the optimal model for application.The results show that:(1) When fitting monthly electricity consumption,the average relative error of temperature prediction method,grey model,autoregressive integrated moving average model and combination forecasting method are 5.44%,3.47%,2.94% and 2.45% respectively;when fitting annual electricity consumption,the average relative errors of temperature prediction method,grey model,autoregressive integrated moving average model,The per capita electricity consumption method and combination forecasting method are only 1.87%,2.22%,1.09%,2.36% and 1.05%,respectively;From the perspective of fitting accuracy,grey model and autoregressive integrated moving average model are relatively ideal;while temperature prediction method and per capita electricity consumption method aren’t quite good;due to the integration of above models,combination forecasting method has the highest accuracy.(2) Results obtained by using combination forecasting method indicate that:In the year of 2019,2020 and 2021,the average growth rate of monthly electricity consumption are 3.65%,4.72% and 4.65% respectively,and the growth rate of annual electricity consumption are 5.12%,5.18% and 5.12% respectively.This conclusion can be used for load management of Guangzhou power grid.(3) It requires more detailed electricity consumption and demographic data to improve the prediction accuracy,and related departments of local government could take some measures such as strengthening the data reporting and communication mechanisms to achieve this goal.

关 键 词：用电量预测 灰色模型 随机季节模型 人均用电量法 组合预测法 

分 类 号：F206[经济管理—国民经济] F062.2