宁夏电力Issue(2)：58-65,8.DOI:10.3969/j.issn.1672-3643.2025.02.009
基于时序图注意力网络与因果推理的配电网故障预测
Fault prediction of distribution networks based on temporal graph attention network and causal inference
摘要
Abstract
Accurate modeling and prediction of fault propagation in distribution networks are vital for ensuring the secure operation of power systems.This paper proposes a fault evolution prediction model based on the temporal graph attention network(TGAT),which integrates graph neural networks with time-series modeling to capture the dynamic patterns of fault propagation.To enhance both prediction accuracy and interpretability,a causal inference module is incorporated to identify key driving factors of fault evolution.Furthermore,a dynamic graph update mechanism and multi-scale feature fusion are introduced,allowing the model to adapt to evolving network topologies.Experimental results demonstrate that the proposed model achieves superior prediction accuracy and robustness under dynamic topolo-gical conditions and significantly improves the identification of critical causal factors.This research offers theoretical support for modeling fault propagation in dynamic distribution networks.关键词
配电网/故障演化预测/时序图注意力网络/因果推理/动态图更新Key words
distribution network/fault evolution prediction/temporal graph attention network/causal inference/dynamic graph update分类
信息技术与安全科学引用本文复制引用
王超,杨雯,聂婉婷,胡斌华,刘品汐..基于时序图注意力网络与因果推理的配电网故障预测[J].宁夏电力,2025,(2):58-65,8.基金项目
国网宁夏电力有限公司科技项目(5229ZW240005) （5229ZW240005）
