电力系统自动化2025，Vol.49Issue(10)：112-122,11.DOI:10.7500/AEPS20240316001
基于改进时空图神经网络的高渗透率有源配电网故障定位
Fault Location for Active Distribution Network with High Penetration Rate Based on Improved Spatio-Temporal Graph Neural Network
摘要
Abstract
Aiming at the problems that there are few studies on fault location for distribution network based on synchronous waveform measurement data,the low utilization of traditional intelligent methods on the network physical topology and waveform data information,and the high proportion of distributed generator access reduce the accuracy of existing methods,a fault location method for active distribution network based on improved spatio-temporal graph neural network is proposed.First,the synchronous waveform measurement data of the distribution network is mapped to the spatio-temporal graph-structured data,and the robustness of the method is improved by combining the physical structure information of the power grid.Then,the spatio-temporal information of the data is maximally utilized by the spatio-temporal fusion graph convolution to extract fault location characteristics and achieve more accurate fault section location with high proportion of distributed generators.Finally,residual connection and gate activation function are introduced to expand the receptive field and reduce the demand for measurement conditions.Simulation results show that the proposed method can locate fault lines with high accuracy under high proportion of distributed generators,various fault conditions and noise interference environments.关键词
配电网/故障定位/分布式电源/波形测量数据/深度学习/图神经网络Key words
distribution network/fault location/distributed generator/waveform measurement data/deep learning/graph neural network引用本文复制引用
黄南天,程铎,蔡国伟..基于改进时空图神经网络的高渗透率有源配电网故障定位[J].电力系统自动化,2025,49(10):112-122,11.基金项目
国家重点研发计划资助项目(2022YFB2404002). This work is supported by National Key R&D Program of China(No.2022YFB2404002). （2022YFB2404002）
