基于稀疏自适应学习的台区用户拓扑结构校验  

Transformer area topology verification method based on sparse adaptive learning

作　　者：冯振宇 沈浚 汪东耀 刘英 温桂平 Feng Zhenyu;Shen Jun;Wang Dongyao;Liu Ying;Wen Guiping(Zhejiang Haining Power Supply Company,State Grid Corporation of China,Haining 314400,Zhejiang,China;School of Information Science and Electronic Engineering,Zhejiang University,Hangzhou 310027,China;Zhejiang Creaway Automation Engineering Co.,Ltd.,Hangzhou 310012,China)

机构地区：[1]国网浙江海宁市供电有限公司,浙江海宁314400 [2]浙江大学信息与电子工程学院,杭州310027 [3]浙江华云信息科技有限公司,杭州310012

出　　处：《电测与仪表》2020年第7期29-34,共6页Electrical Measurement & Instrumentation

基　　金：国家自然科学基金资助项目(61471320);浙江省自然科学基金资助项目(LY17F010009)。

摘　　要：针对低压台区拓扑结构人工校验成本高且准确性不足的问题,提出了基于稀疏自适应学习的台区用户拓扑结构校验方法。基于用电信息系统采集的用电量数据,构建了参数化台区用电量模型,提出了稀疏自适应学习方法自动估计出模型参数。通过阈值检验识别出台户拓扑结构统计错误的用户。采用浙江省海宁地区的用电量数据对该方法的性能进行分析。实验结果表明,该方法具有较好的识别率。在模拟场景中,可以达到100%的查全率和查准率;在真实场景中,可以达到84.8%的查准率和90.7%的查全率。Aiming at the problem of high cost and low accuracy using artificial verification in low-voltage transformer area topology,this paper proposes a new automatic verification method based on sparse adaptive learning.Based on the electricity consumption data of massive users,a parametric electricity consumption model of low-voltage transformer areas was constructed.Then,a sparse adaptive learning algorithm was proposed to automatically estimate the model parameters.Users who do not belong to the transformer area were identified by utilizing a threshold testing.The performance of the proposed method was tested using the electricity consumption data of a certain transformer area in Haining,Zhejiang province.Experimental results showed that the proposed method achieved a good estimation performance.In the simulative cases,the proposed method can achieve 100%accuracy ratio and 100%recall ratio.In the real cases,it can achieve 84.8%accuracy ratio and 90.7%recall ratio.

关 键 词：拓扑结构校验 稀疏学习 低压台区 用电量 参数估计 最小均方误差 

分 类 号：TM715[电气工程—电力系统及自动化]