一种基于双层聚类分析的负荷形态组合识别方法-电网技术.pdfVIP

40 卷 5 期 电 网 技 术 Vol. 40 No. 5 20 16 年5 月 Power System Technology May 2016 文章编号：1000-3673 （20 16）05-1495-07 中图分类号：TM 7 11 文献标志码：A 学科代码：470 40 一种基于双层聚类分析的负荷形态组合识别方法 王星华，陈卓优，彭显刚 （广东工业大学 自动化学院，广东省 广州市 510006 ） A New Combinational Electrical Load Analysis Method Based on Bilayer Clustering Analysis WANG Xinghua, CHEN Zhuoyou, PENG Xiangang (School of Automation, Guangdong University of Technology, Guangzhou 510006, Guangdong Province, China) ABSTRACT: 1Different from traditional electricity customer 0 引言 behavior analysis method, this paper presents a combinational 随着大数据时代的到来，在智能电网的大环境 load profile recognition method with two layers based on improved clustering algorithm. In order to overcome defect of 下，电力公司用电信息采集系统、电力营销系统和 traditional clustering algorithm, sensitive to initial conditions 客户服务信息系统等积累了各行各业海量用电信 and usually leading to local minimum results, selection rules [1] 息 ，且数据之间存在关联性。探索有效的数据挖 for two layers of initial cluster centers are proposed. It changes 掘方法、充分挖掘与分析用户用电行为，对指导用 cluster center selection from random set to purposeful selection. 户合理用电及安排供电规划、高效利用能源具有重 On outer layer, it uses cosine similarity function as its [2-3] 要意义 。目前，现有的研究大多采用聚类分析的 performance evaluation index; on inner layer, Euclidean distance function is used as evaluation index. For the proposed 方法对用户用电轨迹聚类来识别电力用户的用电 method, it is not necessary to normalize load curve vector in [4-7] 模式 。针对基于多种聚类算法、作为掌握用户负