基于深度信念网络的不同行业中长期负荷预测  

Medium-and Long-term Load Forecasting for Different Industries Based on Deep Belief Network

作　　者：张籍 薛儒涛 刘慧 陈艳波[2] 谢东 高晓晶 ZHANG Ji;XUE Rutao;LIU Hui;CHEN Yanbo;XIE Dong;GAO Xiaojing(Economic and Technology Research Institute,State Grid Hubei Electric Power Company,Wuhan 430063,China;State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources,North China Electric Power University,Beijing 102206,China;State Grid Hubei Electric Power Company,Wuhan 430077,China)

机构地区：[1]国网湖北省电力有限公司经济技术研究院,武汉430063 [2]华北电力大学新能源电力系统国家重点实验室,北京102206 [3]国网湖北省电力有限公司,武汉430077

出　　处：《电力系统及其自动化学报》2019年第9期12-19,27,共9页Proceedings of the CSU-EPSA

基　　金：国网湖北省电力公司科技资助项目(521538160012)

摘　　要：为保证不同行业中长期负荷预测的准确性,提出一种基于深度信念网络的不同行业中长期负荷预测方法。首先,采用灰色关联度分析法定量分析各种影响因素对不同行业的影响程度,生成关联度矩阵;然后,基于关联度矩阵,采用模糊C-均值聚类法将不同行业划分为不同的预测类型;其次,针对每种预测类型建立基于深度信念网络的中长期负荷预测模型;最后,采用实际电网数据测试所提方法的精度,结果显示本文方法得到的中长期负荷预测平均误差率与最大误差率分别低于2%与6%,证明了所提方法对中长期负荷预测具有较高的准确性。To ensure the accuracy of medium-and long-term load forecasting for different industries,a medium-and longterm load forecasting method based on deep belief network(DBN)is proposed. First,the grey correlation analysis method is used to quantitatively analyze the impacts of various factors on different industries and generate a correlation degree matrix,based on which different industries are classified into different forecasting types by using the fuzzy Cmeans clustering method. Then,a medium-and long-term load forecasting model based on DBN is established for each forecasting type. Finally,the accuracy of the proposed method is tested by actual power grid data,and results show that the average and maximum error rates obtained using this method are lower than 2% and 6%,respectively,proving that the proposed method has higher accuracy for medium-and long-term load forecasting.

关 键 词：中长期负荷预测 深度学习 深度信念网络 关联度分析 聚类分析 行业分类 

分 类 号：TM715[电气工程—电力系统及自动化]