电力信息与通信技术2025，Vol.23Issue(11)：18-27,10.DOI:10.16543/j.2095-641x.electric.power.ict.2025.11.03
基于多源数据融合的配电网线路转供行为识别方法研究与应用
Research and Application of Feeder Load Transferring Behavior Identification Method Based on Multi-source Data Fusion in Distribution Network
摘要
Abstract
Accurately identifying the load transferring behavior of distribution network is the key foundation for doing a good job in safety maintenance,customer service,line loss management and other businesses.Aiming at the problem that traditional feeder load transferring behavior cannot be identified in full quantity,this paper proposes a method for identifying distribution network load transferring behavior based on multi-source data fusion.Firstly,a pool of triggering events of suspected transferring behavior is established by comprehensively considering the information of feeder switch displacement,fault outage records,line loss status,etc.Secondly,a collection of suspected transferring feeders to be diagnosed is formed by analyzing the interconnection switch status of the feeders triggering the events.Finally,a collection of suspected transferring feeders to be diagnosed is identified by using the transferring behavior checking model based on the integrated learning algorithm of LightGBM,and the real transferring feeders are identified.Finally,using the LightGBM integrated learning algorithmbased calibration model for the suspected transferring feeders to be diagnosed,the real transferring feeders are recognized.Practical cases have proved the effectiveness of the above method.The method has been applied in a province,accurately identifying load transferring behavior and improving the efficiency of line loss,reliability management and other operations.关键词
配电网/转供行为识别/LightGBM集成学习/多源数据融合/线损管理Key words
distribution network/load transferring behavior identification/LightGBM integrated learning/multi-source data fusion/line loss management分类
动力与电气工程引用本文复制引用
陈烨,陈锦铭,赵新冬..基于多源数据融合的配电网线路转供行为识别方法研究与应用[J].电力信息与通信技术,2025,23(11):18-27,10.基金项目
国网江苏省电力有限公司科技项目资助"数字电网精准映射与计算推演技术研究"(J2023121). （J2023121）
