电力信息与通信技术2025，Vol.23Issue(11)：43-50,8.DOI:10.16543/j.2095-641x.electric.power.ict.2025.11.06
面向电网巡检的5G深度指纹定位算法
5G Deep Fingerprint Positioning Algorithm for Power Grid Inspection
摘要
Abstract
In the smart grid,5G based rail mounted inspection robots can replace human labor for intelligent and efficient safety inspections of power grid equipment.While breaking through the limitations of traditional manual inspections,they also place higher demands on the accuracy and stability of indoor positioning algorithms.In response to the problems of low positioning accuracy and poor stability in traditional indoor positioning methods,this paper proposes an improved graph sample and aggregate(GraphSAGE)neural network positioning method.Firstly,the fingerprint data of multiple RF signals are converted into heterogeneous graph data,which is input into the GraphSAGE neural network to obtain the initial positioning result.Then,the weighted K-nearest neighbor algorithm is used to optimize the positioning result.The experimental results demonstrate that the proposed improved GraphSAGE neural network positioning algorithm effectively improves positioning accuracy and has high system stability.关键词
室内定位/多源融合/几何深度学习/图采样与聚合神经网络/加权K近邻Key words
indoor positioning/multi-source fusion/geometric deep learning/graph sampling and aggregation neural network/weighted k-nearest neighbor分类
信息技术与安全科学引用本文复制引用
蔡万升,解鹏,宋曦..面向电网巡检的5G深度指纹定位算法[J].电力信息与通信技术,2025,23(11):43-50,8.基金项目
南瑞集团自筹科技项目"基于RedCap技术的电力5G工业模组及通信终端关键技术研究及应用"(5246DR230033). （5246DR230033）
