基于两阶段聚类的单相电能表质量评价模型研究  

Research on quality evaluation model of single phase electric energy meter based on two-stage clustering

作　　者：夏伟钊 滕欢[1] 曹敏 沈鑫 赵楠 Xia Weizhao;Teng Huan;Cao Min;Shen Xin;Zhao Nan(School of Electrical Engineering,Sichuan University,Chengdu 610065,China;Electric Power Research Institute of Yunnan Power Grid Limited Liability Corporation,Kunming 650217,China)

机构地区：[1]四川大学电气工程学院,成都610000 [2]云南电网有限责任公司电力科学研究院,昆明650217

出　　处：《电测与仪表》2020年第7期120-125,共6页Electrical Measurement & Instrumentation

摘　　要：电能表的检定误差是评价电能表质量的重要指标之一。当前,行业内主要以基本误差限作为合格与不合格的评价指标,缺乏对合格指标的进一步细分。此外,人工处理的工作量巨大。针对这些问题,提出了一种基于两阶段聚类的电能表质量评价模型。首先分析了各种聚类方法的适用范围与A、B、C三个厂家单相电能表的检定数据,选择K-means算法与HDBSCAN(Hierarchical Density-Based Spatial Clustering of Applications with Noise)作为分析方法。随后,在第一阶段聚类采用HDBSCAN确定K-means算法的K值和初始聚类中心,以解决初始聚类中心随机化的问题。在第二阶段聚类采用K-means算法对所有三个厂家的检定数据进行聚类,得出各聚类类别的统计量描述以确定类别的优劣性,并对结果进行验证。最后,计算得出各厂家和各型号的检定数据在不同类别的占比,给出评价模型的一般推广。所提方法可为电网公司单相电能表厂家与型号的选择提供参考。The verification error of the electric energy meter is one of the important indicators for evaluating the power quality.At present,the industry mainly adopts the basic error limit as the evaluation index of pass and fail,and lacks further subdivision.Moreover,the amount of manual processing is huge.Aiming at these problems,a quality evaluation model of electric energy meter based on two-stage clustering is proposed in this paper.Firstly,the application range of various clustering methods and the verification data of single-phase electric energy meter of A,B and C manufacturers are analyzed.K-means and HDBSCAN(hierarchical density-based spatial clustering of applications with noise)are selected as the analysis method.Subsequently,in order to solve the problem of initial clustering center randomization of K-means,HDBSCAN is adopted to determine the K value and the initial clustering center in first stage clustering.Then,K-means is used to cluster the verification data of all three manufacturers in second stage clustering.The statistical description of each cluster category is obtained to determine the advantages and disadvantages of the categories,and the results are verified.Finally,the proportion of the verification data of each manufacturer and model in different category is calculated.The general promotion of the evaluation model is given.The proposed method can provide reference for the selection of single-phase electric energy meter manufacturers and models in power grid companies.

关 键 词：聚类分析 电能计量 HDBSCAN K-MEANS 检定误差 数据挖掘 

分 类 号：TM933[电气工程—电力电子与电力传动] TP391[自动化与计算机技术—计算机应用技术]