改进K-means算法的馈线线损计算  

Calculation and Application of Feed Line Loss Based on Improved K-means Algorithm

作　　者：张海林[1] 李琳 夏传良[1] ZHANG Hai-lin;LI Lin;XIA Chuan-liang(School of Computer Science&Technology,Shandong Jianzhu University,Jinan 250101,China;Material Company of State Grid Shandong Electric Company,Jinan 250001,China)

机构地区：[1]山东建筑大学计算机科学与技术学院,山东济南250101 [2]国网山东省电力公司物资公司,山东济南250001

出　　处：《软件导刊》2019年第12期22-25,29,共5页Software Guide

基　　金：山东省自然科学基金项目(ZR2016FM19)

摘　　要：分析几种主要线损计算方法优缺点及线损分析中数据挖掘算法应用,提出基于线损时域特征指标和改进K-means算法的馈线线损计算方法。充分利用线损信号中的时域信息,获取线损信号中的平均线损率、线损率变异系数、线损率变化趋势等表征线损信号的非平稳特征。使用该算法对区域889条馈线线损进行计算分析,取轮廓系数最大时对应的k值进行聚类分析,经过65次迭代得到8个聚类结果,其中第7类平均线损率高达33.5%,第5类线损率为17.8%,但线损率变化趋势达308。可以进一步对该类馈线上的用电客户负荷曲线进行跟踪分析,确定是否存在窃漏电行为。The advantages and disadvantages of several main calculation methods of line loss and the application of data mining algo?rithms in analysis of line loss is analyzed in this paper.The calculation method of feeder line loss based on time domain characteristic index of line loss and improved K-means algorithm is proposed,which makes full use of time domain information in signal of line loss.The average line loss rate,the line loss rate variation coefficient,and the line loss rate change trend in the signal of line loss were used to characterize the non-stationary characteristics of the line loss signal.The algorithm is used to calculate and analyze the 889 feeder line losses in the region.The corresponding k value is used to cluster and analyze when the contour coefficient is maximum.The eight clustering results are obtained after 65 iterations.The average line loss rate of the 7th class was as high as 33.5%,the line loss rate of 5th class was 17.8,but the line loss rate reached 308.The power consumption curve of the feeders can be further analyzed to determine whether there is any leakage behavior.

关 键 词：馈线线损 改进K-MEANS算法 聚类分析 轮廓系数 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]