基于谱聚类的电力负荷分类  

Power Load Classification Based on Spectral Clustering

作　　者：李天云[1] 李想[2] 刘辉军[1] 王洪涛[1] 

机构地区：[1]东北电力大学,吉林吉林132012 [2]梧州供电局,广西梧州543002

出　　处：《吉林电力》2008年第5期4-6,共3页Jilin Electric Power

摘　　要：在谱聚类研究成果的基础上,提出了一种新的电力负荷聚类方法,该方法能克服目前常用聚类算法的不足,对聚类数据无特殊要求,无须预先确定分类的数目,算法简单无须反复迭代。通过对实际算例的计算分析以及与k-均值方法的比较,体现了其有效性和优越性。A new power load clustering method is proposed based on the study achievement of spectral clustering.Compared with the present classification methods,this method has no special requirements for characteristics of clustering data and pre-determination of classification number.With simple algrithm and no repeated iteration,this method shows efficiency and advantage by analyzing the example calculation and comparing with the k-means method.

关 键 词：谱聚类 电力负荷 k-均值法 负荷建模 相似度函数 

分 类 号：TM714[电气工程—电力系统及自动化] TM743