摘要
针对电力公司目前在低压配电网中缺乏中间馈线连接关系信息以及拓扑变动导致档案“前清后乱”的难题,随着智能量测体系的发展,提出了一种融合用户负荷特征的低压配电网拓扑关系识别方法。首先,借助智能量测体系提供的非介入式负荷辨识结果作为负荷特征数据源,提取用户设备负荷与电压暂态特征时间序列;然后,结合用户负荷辨识结果对动态时间弯曲距离算法进行加权改进,提高相似度计算准确度,实现户-变与相位识别;最后,考虑到用户设备开启的随机性,基于分段聚合近似法对电压特征时间序列进行降维,并基于密度聚类和聚类后簇内电压特征大小排序实现户-线识别。算例充分考虑城市台区和乡村台区2种不同低压配电网结构,实现户-线-变全拓扑识别且用户拓扑准确率超过95%。
The lack of intermediate feeder connection information in the low-voltage distribution network(LVDN)and the problem of archives becoming disordered due to topological changes pose a challenge for power companies.With the development of intelligent measurement systems,this paper proposes a low voltage distribution network topology recognition method that integrates user load characteristics.The algorithm firstly uses the non-intrusive load identification results provided by the intelligent measurement system as a source of load feature data and extracts user device load and voltage transient feature time series.Then,by combining the user load identification results with a weighted improvement of the dynamic time warping(DTW)algorithm,the algorithm improves the accuracy of similarity calculations and achieves household-to-transformer and phase identification.Finally,considering the randomness of user device activation,the voltage feature time series are dimensionally reduced using the piecewise aggregate approximation(PAA)method,and"household-to-line"recognition is achieved based on density clustering and clusteringbased intra-cluster voltage feature size ranking.The case study considers both urban and rural low-voltage distribution network structures and achieves full topology recognition of"household-to-line-to-transformer"with a user topology accuracy of over 95%.
作者
方磊
冯隆基
杨锦成
黄时
张娅楠
FANG Lei;FENG Longji;YANG Jincheng;HUANG Shi;ZHANG Yanan(State Grid Jiangsu Electric Power Co.,Ltd.,Nanjing Power Supply Company,Nanjing 210008,China;Southeast University,Nanjing 210096,China)
出处
《供用电》
北大核心
2024年第11期60-68,共9页
Distribution & Utilization
基金
国家重点研发计划项目(2021YFB2501600)
国网江苏省电力有限公司新技术项目“智能物联电子式开关研发试点应用”(JS2101852)。
关键词
负荷辨识
低压配电网
动态时间弯曲
分段聚合近似法
拓扑识别
load identification
low voltage distribution network(LVDN)
dynamic time warping(DTW)
piecewise aggregate approximation(PAA)method
topology identification
作者简介
方磊(1991-),男,博士,高级工程师,主要从事电能计量、光伏发电、智能车联网及能源互联网技术的研究工作;冯隆基(1981-)男,硕士,高级工程师,主要从事电能计量、营销大数据等研究工作;杨锦成(1993-)男,硕士,工程师,主要从事电能计量、营销大数据等研究工作。
