基于密度峰值的高维电力负荷数据聚类方法  

High⁃dimensional power load data clustering method based on density peak

作　　者：郭晓霞 刘佳易 程昱舒 GUO Xiaoxia;LIU Jiayi;CHENG Yushu(State Grid Shanxi Marketing Service Center,Taiyuan 030000,China)

机构地区：[1]国网山西省电力公司营销服务中心,山西太原030000

出　　处：《电子设计工程》2024年第20期103-106,111,共5页Electronic Design Engineering

基　　金：国网山西省电力公司科技项目资助(52051L20000A)。

摘　　要：由于电力能源应用量的暴增,电力负荷数据体量也逐渐加大,隐藏信息挖掘难度越来越大,对负荷数据处理技术提出了更高的要求,为此提出基于密度峰值的高维电力负荷数据聚类方法。深入剖析电力负荷数据特征,检测并修正其中的异常数据,去除负荷曲线基荷部分,完成负荷数据的预处理。确定电力负荷数据局部密度计算公式,引入密度峰值聚类算法制定高维电力负荷数据聚类程序,执行指定程序即可获得负荷数据聚类结果。实验数据显示,应用提出方法后,DBI指标最小值为0.22,FMI指标最大值为0.96,表明其数据聚类效果更好,证实了提出方法的应用性能较佳。Due to the explosion of power energy applications,the volume of power load data is also gradually increasing,and the difficulty of hidden information mining is increasing,which puts forward higher requirements for load data processing technology.Therefore,a high-dimensional power load data clustering method based on peak density is proposed.Deeply analyze the characteristics of power load data,detect and correct the abnormal data,remove the base load part of the load curve,and complete the load data preprocessing.Determine the local density calculation formula of power load data,introduce the density peak clustering algorithm to develop the high-dimensional power load data clustering program,and the load cluster analysis results can be obtained by executing the specified program.The experimental data shows that after the proposed method is applied,the minimum value of DBI index is 0.22,and the maximum value of FMI index is 0.96,which indicates that its data clustering effect is better,and confirms that the proposed method has better application performance.

关 键 词：高维数据 数据聚类 密度峰值 电力负荷数据 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]