数据驱动的低压配电台区拓扑辨识技术综述  

Overview of Data-driven Topology Identification Techniques for Low-voltage Distribution Substations

作　　者：戚成飞 刘岩 毕超然 王耀宇 李文文 易姝娴 QI Chengfei;LIU Yan;BI Chaoran;WANG Yaoyu;LI Wenwen;YI Shuxian(Marketing Service Center,State Grid Jibei Electric Power Co.,Ltd.,Beijing 100032,China;Electric Power Research Institute,State Grid Jibei Electric Power Co.,Ltd.,Beijing 100045,China)

机构地区：[1]国网冀北电力有限公司营销服务中心,北京100032 [2]国网冀北电力有限公司电力科学研究院,北京100045

出　　处：《电力系统及其自动化学报》2024年第10期127-134,共8页Proceedings of the CSU-EPSA

基　　金：国网冀北电力有限公司科技项目(52018520002U)。

摘　　要：低压配网拓扑结构是新型配电网系统分析所关注的重点领域,高级量测体系建设为低压配网积累大量数据,数据驱动类方法备受关注。根据数据源类型不同,本文将低压配电台区拓扑识别方法分为两类:第1类基于电气量测量数据的机器学习与数据分析方法,包括相关性分析、聚类分析、数学规划和深度学习;第2类基于通信信号数据,利用电力线载波通信,采用调制信号、信号强度检测方法进行拓扑识别。最后,基于对现有方法的对比分析,对今后工作进行了展望,旨在为科研人员和工程技术人员提供参考。The low-voltage(LV)distribution network topology is a key area of concern for the analysis of novel distribution systems.Since the construction of advanced metering infrastructure has accumulated a large amount of data for LV distribution networks,the data-driven methods have attracted much attention.According to the different types of data sources,the topology identification methods for LV distribution substations are classified into two categories in this paper.The first category is based on the machine learning and data analysis methods of electrical measurement data,which includes correlation analysis,cluster analysis,mathematical planning,and deep learning.The second category is based on the communication signal data,and it uses the power line carrier communication and adopts the modulation signal and signal strength detection methods for topology identification.Finally,based on the comparative analysis of the existing methods,the work in the future is prospected with an aim of providing reference for researchers and engineers.

关 键 词：拓扑辨识 低压配电网 高级量测体系 机器学习 电力线载波通信 

分 类 号：TM7[电气工程—电力系统及自动化]