摘要
海上风电场由于风机或线路的投退会引起集电线路拓扑变化,现有基于人工智能算法的集电线路故障区段定位方法,在拓扑变化后定位性能会下降。为此,基于图采样聚合算法构建海上风电场集电线路故障区段定位模型,把线路区段映射为边图模型的节点,各区段故障前后的电流有效值作为节点特征,将故障区段定位问题转化为图节点分类问题。仿真结果表明,所提定位方法充分挖掘了集电线路的拓扑结构特征和电流数值特征,定位准确率高,抗干扰能力强,且有效提高了定位模型对拓扑变化的适应能力。
In offshore wind farms,the topology of collector lines will change due to the input or withdraw of wind turbines or lines.However,the performance of the existing methods for locating the fault segments of collector lines based on artificial intelligence algorithms will degrade after the topological changes.To solve this problem,a fault segment location model of collector lines in offshore wind farms based on the graph sample and aggregate algorithm is constructed.The line segment is mapped as the node of an edge graph model,and the effective values of the pre-and post-fault current in each segment are taken as node characteristics.Therefore,the problem of fault segment location is transformed into a graph node classification problem.Simulation results show that the proposed location method fully excavates the topological structure characteristics of the collector line and the numerical characteristics of current,and it has a high location accuracy and a strong anti-interference capability.In addition,it effectively improves the adaptability of the location model to topological changes.
作者
白通
王慧芳
杨林刚
高玉青
何瑶璐
BAI Tong;WANG Huifang;YANG Lingang;GAO Yuqing;HE Yaolu(School of Electrical Engineering,Zhejiang University,Hangzhou 310027,China;PowerChina Huadong Engineering Corporation Limited,Hangzhou 311122,China)
出处
《电力系统及其自动化学报》
CSCD
北大核心
2024年第10期108-116,共9页
Proceedings of the CSU-EPSA
关键词
故障区段定位
海上风电场
集电线路
图表示学习
拓扑变化
fault segment location
offshore wind farm
collector line
graph representation learning
topological change
