1.国电南瑞科技股份有限公司,江苏省南京市 211106
2.国网电力科学研究院有限公司(南瑞集团有限公司),江苏省南京市 211106
3.电网运行风险防御技术与装备全国重点实验室,江苏省南京市 211106
崔天泽(2000—),男,硕士研究生,主要研究方向:负荷预测、光储充电站运行优化等。E-mail:910421131@qq.com
滕贤亮(1978—),男,通信作者,研究员级高级工程师,博士生导师,主要研究方向:电网调度自动化、电力市场。E-mail:tengxianliang@sgepri.sgcc.com.cn
王俊(1995—),男,工程师,主要研究方向:综合能源系统、电力市场与电力经济、需求响应。E-mail:wangjun18@sgepri.sgcc.com.cn
收稿:2025-05-29,
修回:2025-09-23,
网络首发:2026-08-21,
移动端阅览
崔天泽,滕贤亮,王俊等.基于图卷积-时域融合Transformer的电动汽车充电负荷预测[J/OL].电力系统自动化, http://dx.doi.org/10.7500/AEPS20250529008.
CUI Tianze,TENG Xianliang,WANG Jun,et al.Electric Vehicle Charging Load Forecasting Based on Graph Convolution and Temporal-fusion Transformer[J/OL].Automation of Electric Power Systems, http://dx.doi.org/10.7500/AEPS20250529008.
崔天泽,滕贤亮,王俊等.基于图卷积-时域融合Transformer的电动汽车充电负荷预测[J/OL].电力系统自动化, http://dx.doi.org/10.7500/AEPS20250529008. DOI:
CUI Tianze,TENG Xianliang,WANG Jun,et al.Electric Vehicle Charging Load Forecasting Based on Graph Convolution and Temporal-fusion Transformer[J/OL].Automation of Electric Power Systems, http://dx.doi.org/10.7500/AEPS20250529008. DOI:
在双碳战略背景下,电动汽车规模化接入电网,充电负荷由于其固有的时空随机性呈现出复杂特征,加剧了对电网供需平衡的影响。精准的充电负荷预测是支撑电网安全运行的关键技术之一。针对图卷积方法在提取充电站空间相关信息时拼接外部特征产生的误差问题,提出一种基于图卷积-时域融合Transformer的充电负荷预测方法。首先,为避免外部特征直接参与图卷积,仅使用历史充电负荷数据生成空间邻接关系,并基于自适应图卷积-门控时间卷积网络模块提取空间拓扑信息;然后,基于隐含空间拓扑信息的隐藏层输出和外部特征生成综合时空特征嵌入,并使用时域融合Transformer进一步学习多源特征间的非线性时序依赖与动态演化规律;最后,通过华东某市的真实历史运行数据验证了所提方法的有效性。相比传统图方法,在31个充电站的15 min级充电负荷预测任务中,所提方法的平均绝对误差指标提升超9%。
Under the background of the dual‑carbon strategy
large-scale electric vehicles are connected to the power grid. Owing to its inherent spatiotemporal randomness
the charging load exhibits complex characteristics and aggravates the impact on the supply-demand balance of the power grid. Accurate charging load forecasting is one of the key technologies to support the safe operation of power grids. Aiming at the error caused by concatenating external features when graph convolutional methods extract spatial correlation information among charging stations
this paper proposes a charging load forecasting method based on graph convolution and temporal-fusion Transformer. Firstly
to avoid direct involvement of external features in graph convolution
only historical charging load data are adopted to generate spatial adjacency relationships
and spatial topological information is extracted via an adaptive graph convolution-gated temporal convolutional network module. Secondly
comprehensive spatiotemporal feature embeddings are generated from the hidden-layer outputs containing implicit spatial topological information and external features. A temporal-fusion Transformer is employed to further learn the nonlinear temporal dependencies and dynamic evolution rules among multi-source features. Finally
the effectiveness of the proposed method is verified using real historical operational data from a city in East China. Compared with conventional graph-based methods
the mean absolute error of the proposed method is improved by more than 9% in the 15‑minute‑interval charging load forecasting task for 31 charging stations.
刘志强 , 张谦 , 朱熠 , 等 . 计及车-路-站-网融合的电动汽车充电负荷时空分布预测 [J]. 电力系统自动化 , 2022 , 46 ( 12 ): 36 - 45 .
Liu Zhiqiang , Zhang Qian , Zhu Yi , et al . Spatial-temporal distribution prediction of charging loads for electric vehicles considering vehicle-road-station-grid integration [J]. Automation of Electric Power Systems , 2022 , 46 ( 12 ): 36 - 45 .
国家统计局 . 中华人民共和国2024年国民经济和社会发展统计公报 .[EB/OL].[ 2025-05-14 ]. https://www.stats.gov.cn/sj/zxfb/202502/t20250228_1958817.html https://www.stats.gov.cn/sj/zxfb/202502/t20250228_1958817.html .
National Bureau of Statistics of China . Statistical communiqué of the People’s Republic of China on the 2024 national economic and social development [EB/OL]. [ 2025-05-14 ]. https://www.stats.gov.cn/sj/zxfb/202502/t20250228_1958817.html https://www.stats.gov.cn/sj/zxfb/202502/t20250228_1958817.html .
毛建斌 , 杨少兵 , 杨湘彦 , 等 . 基于有限混合模型的规模化电动汽车充电负荷在线预测方法 [J]. 电网技术 , 2025 , 49 ( 5 ): 1931 - 1940 .
Mao Jianbin , Yang Shaobing , Yang Xiangyan , et al . Charging load online forecasting method based on finite mixture model for large-scale electric vehicles [J]. Power System Technology , 2025 , 49 ( 5 ): 1931 - 1940 .
张美霞 , 孙铨杰 , 杨秀 . 考虑多源信息实时交互和用户后悔心理的电动汽车充电负荷预测 [J]. 电网技术 , 2022 , 46 ( 2 ): 632 - 645 .
Zhang Meixia , Sun Quanjie , Yang Xiu . Electric vehicle charging load prediction considering multi-source information real-time interaction and user regret psychology [J]. Power System Technology , 2022 , 46 ( 2 ): 632 - 645 .
黄一修 , 肖仕武 . 考虑气象因素的高速公路服务区电动汽车充电负荷预测 [J]. 电力自动化设备 , 2025 , 45 ( 1 ): 192 - 199 .
Huang Yixiu , Xiao Shiwu . Forecasting of electric vehicle charging load in highway service areas considering meteorological factors [J]. Electric Power Automation Equipment , 2025 , 45 ( 1 ): 192 - 199 .
锁军 , 李龙 , 贺瀚青 , 等 . 考虑交通路况的电动汽车充电负荷预测 [J]. 电网与清洁能源 , 2022 , 38 ( 10 ): 141 - 147 .
Suo Jun , Li Long , He Hanqing , et al . Load forecasting of electric vehicle charging considering traffic conditions [J]. Power System and Clean Energy , 2022 , 38 ( 10 ): 141 - 147 .
张琳娟 , 许长清 , 王利利 , 等 . 基于OD矩阵的电动汽车充电负荷时空分布预测 [J]. 电力系统保护与控制 , 2021 , 49 ( 20 ): 82 - 91 .
Zhang Linjuan , Xu Changqing , Wang Lili , et al . OD matrix based spatiotemporal distribution of EV charging load prediction [J]. Power System Protection and Control , 2021 , 49 ( 20 ): 82 - 91 .
李宏伟 , 宋玉峰 , 李帅兵 , 等 . 基于ArcGIS路网结构与交通拥挤度分析的电动汽车充电负荷预测方法 [J]. 电网技术 , 2025 , 49 ( 5 ): 1920 - 1930 .
Li Hongwei , Song Yufeng , Li Shuaibing , et al . Electric vehicle charging load prediction based on ArcGIS road network structure and traffic congestion analysis [J]. Power System Technology , 2025 , 49 ( 5 ): 1920 - 1930 .
Zhang J , Yan J , Liu Y Q , et al . Daily electric vehicle charging load profiles considering demographics of vehicle users [J]. Applied Energy , 2020 , 274 : 115063 .
Yi L , Ting X T , Song W , et al . The load forecasting of charging stations based on support vector regression [C]// 5th International Conference on Power and Energy Technology (ICPET) , 2023 : 991 - 995 .
Zhang X , Chan K W , Li H R , et al . Deep-learning-based probabilistic forecasting of electric vehicle charging load with a novel queuing model [J]. IEEE Transactions on Cybernetics , 2021 , 51 ( 6 ): 3157 - 3170 .
林祥 , 张浩 , 马玉立 , 等 . 基于改进LSTM神经网络的电动汽车充电负荷预测 [J]. 现代电子技术 , 2024 , 47 ( 6 ): 97 - 101 .
Lin Xiang , Zhang Hao , Ma Yuli , et al . Electric vehicle charging load prediction based on improved LSTM neural network [J]. Modern Electronics Technique , 2024 , 47 ( 6 ): 97 - 101 .
周喆 , 黄婧杰 , 周任军 , 等 . 考虑关键气象因素的时间卷积网络充电桩负荷预测 [J]. 电力系统及其自动化学报 , 2023 , 35 ( 5 ): 28 - 36 .
Zhou Zhe , Huang Jingjie , Zhou Renjun , et al . Charging pile load prediction using temporal convolutional network with consideration of key meteorological factors [J]. Proceedings of the CSU-EPSA , 2023 , 35 ( 5 ): 28 - 36 .
Huang X S , Wu D , Boulet B . MetaProbformer for charging load probabilistic forecasting of electric vehicle charging stations [J]. IEEE Transactions on Intelligent Transportation Systems , 2023 , 24 ( 10 ): 10445 - 10455 .
胡博 , 张鹏飞 , 黄恩泽 , 等 . 基于图WaveNet的电动汽车充电负荷预测 [J]. 电力系统自动化 , 2022 , 46 ( 16 ): 207 - 213 .
Hu Bo , Zhang Pengfei , Huang Enze , et al . Graph WaveNet based charging load forecasting of electric vehicle [J]. Automation of Electric Power Systems , 2022 , 46 ( 16 ): 207 - 213 .
张延宇 , 张智铭 , 刘春阳 , 等 . 基于动态自适应图神经网络的电动汽车充电负荷预测 [J]. 电力系统自动化 , 2024 , 48 ( 7 ): 86 - 93 .
Zhang Yanyu , Zhang Zhiming , Liu Chunyang , et al . Electric vehicle charging load prediction based on dynamic adaptive graph neural network [J]. Automation of Electric Power Systems , 2024 , 48 ( 7 ): 86 - 93 .
Joon K , Kyeom K . Spatial-temporal graph convolutional-based recurrent network for electric vehicle charging stations demand forecasting in energy market [J]. IEEE Transactions on Smart Grid , 2024 , 15 ( 4 ): 3979 - 3993 .
Xie T , Yun X Y , Zhang G , et al . Charging station cluster load prediction: Spatiotemporal multi-graph fusion technology [J]. Renewable and Sustainable Energy Reviews , 2024 , 206 : 114855 .
Shi J K , Zhang W G , Bao Y , et al . Load forecasting of electric vehicle charging stations: attention based spatiotemporal multi–graph convolutional networks [J]. IEEE Transactions on Smart Grid , 2024 , 15 ( 3 ): 3016 - 3027 .
顾玮 , 段敬 , 张栋 , 等 . 基于数据与模型联合驱动的电动汽车充电负荷时空分布预测 [J]. 计算机科学 , 2024 , 51 ( 增刊2 ): 1032 - 1037 .
Gu Wei , Duan Jing , Zhang Dong , et al . Spatial-temporal distribution prediction of electric vehicle charging load based on data and model joint drive [J]. Computer Science , 2024 , 51 ( S2 ): 1032 - 1037 .
You L L , Chen Q Y , Qu H H , et al . FMGCN: federated meta learning-augmented graph convolutional network for EV charging demand forecasting [J]. IEEE Internet of Things Journal , 2024 , 11 ( 14 ): 24452 - 24466 .
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