基于深度特征挖掘的低压配电网相序精准辨识方法  

Precise identification method of low voltage distribution network phase sequence based on deep feature mining

作　　者：严勤 汤克艰 黄莉 俞林刚 伍栋文 YAN Qin;TANG Kejian;HUANG Li;YU Lingang;WU Dongwen(State Grid Jiangxi Electric Power Co.,Ltd.,Nanchang 330096,China;Southeast University,School of Electrical Engineering,Nanjing 210096,China;State Grid Jiangxi Power Supply Service Management Center,Nanchang 330096,China)

机构地区：[1]国网江西省电力有限公司,江西南昌330096 [2]东南大学电气工程学院,江苏南京210096 [3]国网江西省电力有限公司供电服务管理中心,江西南昌330096

出　　处：《供用电》2022年第6期40-47,共8页Distribution & Utilization

基　　金：国家重点研发计划科技助力经济2020专项(SQ2020YF F0426410);国网江西省电力公司科技项目“低压台区运行全状态感知关键技术研究及应用”。

摘　　要：低压用户相序关系是低压配电网三相不平衡治理、可开放容量评估的重要数据基础。目前,相别档案数据错乱严重,主要依靠现场校验进行数据治理,成本高、准确度低且前清后乱。针对这一问题,提出了一种基于深度特征挖掘的低压配电网相别在线辨识方法。首先,基于全时序差分算法与特征显著性分析得到电压差高维特征,并应用主成分分析法进行降维分析,形成低维、高信息深度特征。然后,采用K-means算法对电压特征进行聚类实现一次辨识,随后针对与配电变压器空间距离近导致的易错辨相别用户,提出一种基于电压与功率特征组合的二次优化方法。最后,以南昌市5个实际低压配电网的共计728个用户为例进行算法验证,结果表明所提方法可有效解决台区内电压相似性高场景对相别辨识模型的制约,大幅提升了辨识的准确性,具有较高的工程推广应用价值。Phase sequence relationship of low voltage users is an important data basis for three-phase unbalance management and open capacity evaluation of low voltage distribution network.At present,he data of different archives are seriously disordered,so the data management is mainly based on the on-site verification,which is of high cost,low accuracy and chaos before and after clearing.To solve this difficult problem,an online identification method of low-voltage distribution network identification based on data-driven is proposed in this paper.Firstly,the high-dimensional features of voltage difference are obtained based on full time series difference algorithm and feature significance analysis.Then,the principal component analysis was used to reduce the dimensions of high-dimensional voltage features to form the characteristics of low-dimensional and high information density.We clustered the voltage characteristics to realize the primary identification based on K-means algorithm.Then,aiming at the error prone users caused by the close spatial distance from the distribution transformer,a secondary optimization method based on the combination of voltage and power characteristics is proposed.Finally,728 users of five actual lowvoltage distribution networks in Nanchang are taken as examples to verify the algorithm.The results show that the proposed method can effectively solve the restriction of identification model due to the high voltage similarity in the control stations passing through the stations,greatly improve the accuracy of identification,and has the feasibility of engineering promotion.

关 键 词：低压配电网 相别辨识 电压特征 差分构造 主成分分析 

分 类 号：TM93[电气工程—电力电子与电力传动]