摘要
电力负荷曲线聚类是配用电大数据挖掘的基础。针对常规K-medoids聚类算法的不足,利用CH指标和启发式方法对常规K-medoids聚类算法进行了改进,实验结果表明,改进的K-medoids聚类算法具有更高的聚类有效性,克服了传统K-medoids聚类算法由用户指定簇族数和随机选取初始质心的问题。为方便分析对22个典型用户年负荷曲线进行归一化处理,并分别采用行业性质、常规K-medoids聚类算法和改进K-medoids聚类算法对经过归一化处理的22个典型用户年负荷曲线进行分类,实验结果表明,根据年负荷曲线特性对负荷曲线进行分类相较于按行业性质对负荷曲线进行分类有更好的效果,而且改进K-medoids聚类算法相较于常规K-medoids聚类算法,能够更客观、准确地对负荷曲线进行有效分类,改进K-medoids聚类算法在电力负荷大数据挖掘具有良好的应用前景。
Load curves clustering is a basic task for big data mining in electricity consumption database.To overcome the shortcomings of traditional K-medoids clustering algorithm,the traditional K-medoids clustering algorithm is improved by using the CH indicator and the heuristic method.The experimental results show that the improved K-medoids clustering algorithm has higher clustering effectiveness,and overcomes the problem of the traditional algorithm of user-specified cluster number and random selection of initial centroids.In order to facilitate the analysis,22 typical user annual load curves were normalized,and they were classified by industry nature,traditional K-medoids clustering algorithm and improved K-medoids clustering algorithm.The experimental results show that the classification of load curves according to the characteristics of the annual load curve is better than the industry nature.Moreover,the improved K-medoids clustering algorithm can effectively classify the load curves more objectively and accurately than the traditional K-medoids clustering algorithm,which proves a good application prospect for the improved K-medoids clustering algorithm in power load big data mining.
作者
王华勇
韩松
肖孝天
杨超
WANG Huayong;HAN Song;XIAO Xiaotian;YANG Chao(School of Electrical Engineering,Guizhou University,Guiyang 550025,China;Guian Power Supply Bureau,Guizhou Power Grid Co.,Ltd,Guiyang 550003,China)
出处
《电力科学与工程》
2019年第2期38-43,共6页
Electric Power Science and Engineering
基金
贵州省科技厅联合资金项目(20157635)
作者简介
王华勇(1994—),男,硕士研究生,主要研究方向为数据挖掘及配电网规划与运行;韩松(1978—),男,博士,教授,主要研究方向为交直流电力系统动态分析,新型电力电子装备以及配电网规划与运行等;肖孝天(1989—),男,硕士,主要研究方向为地区电网调控;杨超(1971—),女,副教授,主要研究方向为电力市场。
