基于K-means聚类算法和BP神经网络的代理购电量预测模型研究  

Research on Electricity Prediction Model of Power Purchasing Agent Based on K-means Clustering Algorithm and BP Neural Network

作　　者：于志诚 穆士才 梁晔 李镓辰 林华 陈己宸 金鑫 YU Zhicheng;MU Shicai;LIANG Ye;LI Jiachen;LIN Hua;CHEN Jichen;JIN Xin(State Grid Beijing Chaoyang Power Supply Company,Beijing 100020,China;State Grid Beijing Customer Service Center,Beijing 100010,China;State Grid Beijing Electric Power Company,Beijing 100032,China)

机构地区：[1]国网北京朝阳供电公司,北京100020 [2]国网北京客服中心,北京100010 [3]国网北京市电力公司,北京100032

出　　处：《湖南电力》2025年第1期68-72,共5页Hunan Electric Power

摘　　要：通过对某地区代理购电用户的深入画像分析,研究不同因素对代理购电用户电量的影响;通过聚类算法实现用户群体的分类;通过神经网络算法将纵向时序电量和横向影响因素纳入预测公式,针对不同聚类簇构建符合其特征的预测模型;最后将模型整合,实现对整体电量的高准确率预测。Through the in-depth portrait analysis of power purchasing agents in a certain region,the influence of different factors on the power consumption of power purchasing agents is studied,and the classification of user groups is realized through the clustering algorithm.Through the neural network algorithm,the longitudinal time series electricity and horizontal influencing factors are incorporated into the prediction formula,and the prediction models conforming to the characteristics of different clusters are constructed.Finally,the models are integrated to achieve high accuracy prediction of the overall electricity.

关 键 词：代理购电 电量预测 聚类算法 神经网络 画像分析 

分 类 号：TM714[电气工程—电力系统及自动化]