基于改进K-means聚类算法的行业用电画像  

Industry Power Profile Based on Improved K-means Clustering Algorithm

作　　者：徐硕 莫文雄[1] 栾乐 童锐 刘田 XU Shuo;MO Wenxiong;LUAN Le;TONG Rui;LIU Tian(Guangzhou Power Supply,Guangdong Power Grid Co.,Ltd.,Guangzhou 510000,China)

机构地区：[1]广东电网有限公司广州供电局,广东广州510000

出　　处：《电工技术》2021年第20期123-126,共4页Electric Engineering

基　　金：中国南方电网有限责任公司科技项目(编号080037KK52180050(GZHKJXM20180068))。

摘　　要：随着大数据研究的深入以及配电网自动化的建设,电力大数据日益得到重视。如何基于配电网大数据实现电网用户感知,支撑配电网相关业务,是当前电力公司关注的重点问题。首先对用户年行度数据进行预处理,通过改进K-means算法对用户用电数据进行聚类;随后结合用户所属行业信息,从用电特征对行业用电进行画像;最后基于实际数据实现行业用电画像,验证方法的有效性。With the continuous deepening of big data research and the construction of distribution network automation,power big data is getting more and more attention.How to realize the perception of power grid users based on the big data of the distribution network and support the related business of the distribution network is the key issue of the current power companies.Firstly,the user's annual travel data is preprocessed,and the user's electricity consumption data is clustered through the improved K-means algorithm.Then combined with the user's industry information,the industry power consumption profile is carried out from the power consumption characteristics.Finally,the industry power consumption profile is realized based on the actual data to verify the effectiveness of the method.

关 键 词：行业用电画像 改进K-means聚类法 DBI指标 数据清洗 

分 类 号：TM933[电气工程—电力电子与电力传动]