1.国家电网有限公司国家电力调度控制中心,北京市 100031
2.清华大学电机工程与应用电子技术系,北京市 100084
3.电网安全全国重点实验室(中国电力科学研究院有限公司),北京市 100192
刘金波(1975—),男,硕士,教授级高级工程师,主要研究方向:电力系统调度自动化。E-mail:jinbo-liu@sgcc.com.cn
严亚勤(1978—),男,硕士,教授级高级工程师,主要研究方向:电力系统调度自动化。E-mail:yan-yaqin@sgcc.com.cn
郭庆来(1979—),男,通信作者,教授,博士生导师,主要研究方向:能量管理、电压控制和信息物理系统。E-mail:guoqinglai@tsinghua.edu.cn
收稿:2025-08-31,
修回:2026-07-14,
网络首发:2026-08-21,
移动端阅览
刘金波,严亚勤,郭庆来等.基于图Transformer的电力系统潮流计算方法[J/OL].电力系统自动化, http://dx.doi.org/10.7500/AEPS20250831014.
LIU Jinbo,YAN Yaqin,GUO Qinglai,et al.Power Flow Calculation Method for Power Systems Based on Graph Transformer[J/OL].Automation of Electric Power Systems, http://dx.doi.org/10.7500/AEPS20250831014.
刘金波,严亚勤,郭庆来等.基于图Transformer的电力系统潮流计算方法[J/OL].电力系统自动化, http://dx.doi.org/10.7500/AEPS20250831014. DOI:
LIU Jinbo,YAN Yaqin,GUO Qinglai,et al.Power Flow Calculation Method for Power Systems Based on Graph Transformer[J/OL].Automation of Electric Power Systems, http://dx.doi.org/10.7500/AEPS20250831014. DOI:
随着新型电力系统加速构建,电网规模持续扩大,系统运行方式不确定性和复杂性特征日益凸显,传统数值迭代潮流求解方法应用于大规模电网时,面临收敛困难与计算效率不足的双重挑战。科学智能(AI4S)技术为数据驱动模式的潮流算法提供了新的技术路线,但现有的图神经网络(GNN)方法依赖局部邻域聚合,难以有效捕捉远距离节点间的电气耦合对功率分配和电压分布的影响,在大电网分析计算中应用效果受限。针对以上问题,文中提出一种融合GNN与Transformer的潮流计算方法,引入随机游走位置编码以强化节点拓扑位置感知能力,利用自注意力机制实现全局节点信息的高效交互,结合潮流方程物理约束、幂平均损失函数及数据增强策略优化模型性能。多个场景的实验结果表明,所提方法在计算精度、速度和迁移泛化方面均具有优势,并在某省验证了其工程有效性。
As the construction of new power systems accelerates
the scale of power grids continues to expand
and the uncertainty and complexity of system operation modes become increasingly prominent. When applied to large-scale power grids
traditional numerical iterative power flow solution methods face the dual challenges of convergence difficulty and insufficient computational efficiency. Artificial intelligence for science (AI4S) technology provides a new technical route for data-driven power flow algorithms. However
existing graph neural network (GNN) methods rely on local neighborhood aggregation
making it difficult to effectively capture the impact of electrical coupling between distant nodes on power allocation and voltage distribution
which limits their application in large-scale power grid analysis and computation. To address these issues
this paper proposes a power flow calculation method that integrates a graph neural network with Transformer. A random walk positional encoding is introduced to enhance the perception of nodal topological positions
and a self-attention mechanism is employed to achieve efficient interaction of global node information. The model performance is further optimized by incorporating physical constraints of power flow equations
a power-mean loss function
and a data augmentation strategy. Experimental results under multiple scenarios demonstrate that the proposed method has advantages in computational accuracy
speed
and transfer generalization
and its engineering effectiveness is verified in a certain province.
周勤勇 , 郝绍煦 , 董武 , 等 . “十五五”电网规划安全稳定分析关键问题 [J]. 中国电力 , 2025 , 58 ( 9 ): 138 - 147 .
Zhou Qinyong , Hao Shaoxu , Dong Wu , et al . Key issues in security and stability analysis for the “15th five-year” power grid planning [J]. Electric Power , 2025 , 58 ( 9 ): 138 - 147 .
陈国平 , 李明节 , 许涛 . 特高压交直流电网系统保护及其关键技术 [J]. 电力系统自动化 , 2018 , 42 ( 22 ): 2 - 10 .
Chen Guoping , Li Mingjie , Xu Tao . System protection and its key technologies of UHV AC and DC power grid [J]. Automation of Electric Power Systems , 2018 , 42 ( 22 ): 2 - 10 .
韩英铎 , 陆超 , 宋文超 , 等 . 电力系统在线安全稳定计算分析技术综述与展望 [J]. 中国电机工程学报 , 2024 , 44 ( 17 ): 6733 - 6761 .
Han Yingduo , Lu Chao , Song Wenchao , et al . Review and prospects of online calculation and analysis techniques for power system security and stability [J]. Proceedings of the CSEE , 2024 , 44 ( 17 ): 6733 - 6761 .
段师琪 , 余娟 , 杨知方 , 等 . 基于 N / N- 1潮流内嵌图卷积神经网络的电网运行方式智能调整 [J]. 电工技术学报 , 2025 , 40 ( 19 ): 6115 - 6130 .
Duan Shiqi , Yu Juan , Yang Zhifang , et al . Intelligent adjustment method of power grid operation mode based on N / N- 1 power flow embedded graph convolutional neural network [J]. Transactions of China Electrotechnical Society , 2025 , 40 ( 19 ): 6115 - 6130 .
王一迪 , 李立新 , 於益军 , 等 . 基于深度强化学习的电力系统安全校正控制 [J]. 电力系统自动化 , 2023 , 47 ( 12 ): 121 - 129 .
Wang Yidi , Li Lixin , Yu Yijun , et al . Power system security correction control based on deep reinforcement learning [J]. Automation of Electric Power Systems , 2023 , 47 ( 12 ): 121 - 129 .
刘金波 , 宋旭日 , 杨楠 , 等 . 深度强化学习在电网实时计划编排中的应用 [J]. 电力系统自动化 , 2023 , 47 ( 14 ): 157 - 166 .
Liu Jinbo , Song Xuri , Yang Nan , et al . Application of deep reinforcement learning in real-time plan scheduling of power grid [J]. Automation of Electric Power Systems , 2023 , 47 ( 14 ): 157 - 166 .
牛奎烨 , 刘金波 , 汤奕 , 等 . 计及极端气象事件的年度系统运行场景生成方法 [J]. 电网技术 , 2024 , 48 ( 10 ): 3992 - 4006 .
Niu Kuiye , Liu Jinbo , Tang Yi , et al . A method for generating annual system operating scenarios considering extreme meteorological events [J]. Power System Technology , 2024 , 48 ( 10 ): 3992 - 4006 .
冯卓诚 , 姜彤 , 万凯遥 , 等 . 基于节点电压方程拆解的交替迭代潮流算法 [J]. 电网技术 , 2023 , 47 ( 4 ): 1675 - 1681 .
Feng Zhuocheng , Jiang Tong , Wan Kaiyao , et al . An alternating iterative power flow algorithm based on bus voltage equation decomposition [J]. Power System Technology , 2023 , 47 ( 4 ): 1675 - 1681 .
王甜婧 , 汤涌 , 郭强 , 等 . 基于知识经验和深度强化学习的大电网潮流计算收敛自动调整方法 [J]. 中国电机工程学报 , 2020 , 40 ( 8 ): 2396 - 2406 .
Wang Tianjing , Tang Yong , Guo Qiang , et al . Automatic adjustment method of power flow calculation convergence for large-scale power grid based on knowledge experience and deep reinforcement learning [J]. Proceedings of the CSEE , 2020 , 40 ( 8 ): 2396 - 2406 .
王毅 , 李可萌 , 曹国芳 , 等 . 复杂故障下静态安全分析的部分因子化修正改进方法 [J]. 电力系统自动化 , 2026 , 50 ( 3 ): 167 - 179 .
Wang Yi , Li Kemeng , Cao Guofang , et al . Improved partial refactorization method for static security analysis under complex faults [J]. Automation of Electric Power Systems , 2026 , 50 ( 3 ): 167 - 179 .
Li Xue , Li Fangxing , Yuan Haoyu , et al . GPU-based fast decoupled power flow with preconditioned iterative solver and inexact Newton method [J]. IEEE Transactions on Power Systems , 2017 , 32 ( 4 ): 2695 - 2703 .
郭健 , 周京阳 , 李强 , 等 . 加权消去树建模方法及其在电网计算矩阵排序性能评价中的应用 [J]. 电网技术 , 2018 , 42 ( 4 ): 1316 - 1321 .
Guo Jian , Zhou Jingyang , Li Qiang , et al . Weighted elimination tree modeling method and its application in power grid matrix ordering performance evaluation [J]. Power System Technology , 2018 , 42 ( 4 ): 1316 - 1321 .
张沛 , 杨马婧 , 张放 , 等 . 图数据库与图计算在电力系统调度运行应用综述 [J]. 中国电力 , 2025 , 58 ( 3 ): 119 - 131 .
Zhang Pei , Yang Majing , Zhang Fang , et al . Review of the application of graph database and graph computing in power system dispatch and operation [J]. Electric Power , 2025 , 58 ( 3 ): 119 - 131 .
马智刚 , 卫志农 , 陈胜 , 等 . 基于图计算的交直流混合配电网优化调度 [J]. 电力系统自动化 , 2023 , 47 ( 18 ): 161 - 170 .
Ma Zhigang , Wei Zhinong , Chen Sheng , et al . Optimal dispatch of AC/DC hybrid distribution network based on graph computing [J]. Automation of Electric Power Systems , 2023 , 47 ( 18 ): 161 - 170 .
李鹏 , 黄文琦 , 王鑫 , 等 . 数据与知识联合驱动的人工智能方法在电力调度中的应用综述 [J]. 电力系统自动化 , 2024 , 48 ( 1 ): 160 - 175 .
Li Peng , Huang Wenqi , Wang Xin , et al . Review on application of combined data-knowledge-driven artificial intelligence methods in power dispatching [J]. Automation of Electric Power Systems , 2024 , 48 ( 1 ): 160 - 175 .
鄂维南 . AI助力打造科学研究新范式 [J]. 中国科学院院刊 , 2024 , 39 ( 1 ): 10 - 16 .
E Weinan . AI helps to establish a new paradigm for scientific research [J]. Bulletin of Chinese Academy of Sciences , 2024 , 39 ( 1 ): 10 - 16 .
Vyakaranam B , Mahapatra K , Li X Y , et al . Novel data-driven distributed learning framework for solving AC power flow for large interconnected systems [J]. IEEE Open Access Journal of Power and Energy , 2021 , 8 : 281 - 292 .
Gao M , Yu J , Yang Z , et al . Physics embedded graph convolution neural network for power flow calculation considering uncertain injections and topology [J]. IEEE Transactions on Neural Networks and Learning Systems , 2024 , 35 ( 11 ): 15467 - 15478 .
战鹏祥 , 黄飞虎 , 廖思睿 , 等 . 基于物理机理引导的数据驱动潮流计算方法 [J]. 电网技术 , 2024 , 48 ( 12 ): 5034 - 5045 .
Zhan Pengxiang , Huang Feihu , Liao Sirui , et al . Data-driven power flow calculation method guided by physical mechanism [J]. Power System Technology , 2024 , 48 ( 12 ): 5034 - 5045 .
Chen Qifan , Bu Siqi , Wang Huaiyuan , et al . Real-time multi-stability risk assessment and visualization of power systems: a graph neural network-based method [J]. IEEE Transactions on Power Systems , 2025 , 40 ( 4 ): 2955 - 2968 .
Lin N , Orfanoudakis S , Cardenas O N , et al . Power flow net: power flow approximation using message passing graph neural networks [J]. International Journal of Electrical Power and Energy Systems , 2024 , 160 , 110112 .
Dwivedi V P , Luu A T , Laurent T , et al . Graph neural networks with learnable structural and positional representations [C]// International Conference on Learning Representations , 2022 .
Alon U , Yahav E . On the bottleneck of graph neural networks and its practical implications [C]// International Conference on Learning Representations , 2021 .
Vaswani A , Shazeer N , Parmar N , et al . Attention is all you need [C]// Advances in Neural Information Processing Systems , 2017 : 5998 - 6008 .
Wu Z H , Pan S R , Chen F W , et al . A comprehensive survey on graph neural networks [J]. IEEE Transactions on Neural Networks and Learning Systems , 2021 , 32 ( 1 ): 4 - 24 .
刘金波 , 郭庆来 , 张靖 , 等 . 机理数据融合驱动的图神经网络状态估计方法 [J]. 电力系统自动化 , 2026 , 50 ( 5 ): 173 - 183 .
Liu Jinbo , Guo Qinglai , Zhang Jing , et al . Hybrid mechanism-data-driven state estimation method based on graph neural network [J]. Automation of Electric Power Systems , 2026 , 50 ( 5 ): 173 - 183 .
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 .
Huang Nantian , Cheng Duo , Cai Guowei . Fault location for active distribution network with high penetration rate based on improved spatio-temporal graph neural network [J]. Automation of Electric Power Systems , 2025 , 49 ( 10 ): 112 - 122 .
孙永辉 , 孟雲帆 , 葛磊蛟 , 等 . 人工智能赋能微电网运行优化的应用及展望 [J]. 高电压技术 , 2023 , 49 ( 6 ): 2239 - 2252 .
Sun Yonghui , Meng Yunfan , Ge Leijiao , et al . Application and prospect of microgrid operation optimization enabled by artificial intelligence [J]. High Voltage Engineering , 2023 , 49 ( 6 ): 2239 - 2252 .
Yu Q , Li Z Y . Correlated load forecasting in active distribution networks using spatial-temporal synchronous graph convolutional networks [J]. IET Energy Systems Integration , 2021 , 3 ( 3 ): 355 - 366 .
杨梅 , 刘俊勇 , 刘挺坚 , 等 . 节点图和边图切换卷积驱动的快速静态安全分析方法 [J]. 电网技术 , 2021 , 45 ( 6 ): 2070 - 2080 .
Yang Mei , Liu Junyong , Liu Tingjian , et al . Switching convolution of node graph and line graph-driven method for fast static security analysis [J]. Power System Technology , 2021 , 45 ( 6 ): 2070 - 2080 .
Liao W , Bak-Jensen B , Pillai R J , et al . A review of graph neural networks and their applications in power systems [J]. Journal of Modern Power Systems and Clean Energy , 2022 , 10 ( 2 ): 345 - 360 .
0
浏览量
0
下载量
0
CSCD
关联资源
相关文章
相关作者
相关机构
京公网安备11010802024621