1.华北电力大学 电气与电子工程学院,北京 102206
2.中国电力科学研究院有限公司,北京 100192
李彬(1983),男,副教授,博士,从事电力通信、电气信息技术、电算协同优化调度等方面的研究,direfish@163.com;
翟振杨*(2002),男,硕士生,从事虚拟电厂运行优化、电算协同优化调度等方面的研究,2377548954@qq.com;
祁泽源(2003),男,硕士生,从事电力需求响应、人工智能等方面的研究,871046930@qq.com;
李淑静(2004),女,硕士生,从事大数据与人工智能、电网供需互动等方面的研究,15930205938@163.com;
陈宋宋(1987),男,高级工程师,硕士,从事能效与智能用电技术、电算协同优化调度等方面的研究,chensongsong2010@163.com;
徐靖(1994),男,工程师,博士,从事智能用电、电力电量分析预测、数据中心用能优化等方面的研究,sg_ydjs@163.com。
收稿:2026-06-23,
修回:2026-07-28,
网络首发:2026-08-04,
纸质出版:2026-08-25
移动端阅览
李彬,翟振杨,祁泽源等.考虑算力弹性伸缩的智算中心与电网协同优化运行关键技术综述[J].综合智慧能源,2026,48(08):44-56.
LI Bin,ZHAI Zhenyang,QI Zeyuan,et al.Review of key technologies for coordinated optimal operation of intelligent computing centers and power grids considering elastic scaling of computing power[J].BLASTING,2026,48(08):44-56.
李彬,翟振杨,祁泽源等.考虑算力弹性伸缩的智算中心与电网协同优化运行关键技术综述[J].综合智慧能源,2026,48(08):44-56. DOI: 10.3969/j.issn.2097-0706.2026.08.004.
LI Bin,ZHAI Zhenyang,QI Zeyuan,et al.Review of key technologies for coordinated optimal operation of intelligent computing centers and power grids considering elastic scaling of computing power[J].BLASTING,2026,48(08):44-56. DOI: 10.3969/j.issn.2097-0706.2026.08.004.
针对智算中心作为新型电力系统中高功率密度、高波动关键负荷所引发的协同运行问题,沿着“任务—资源—能耗—电网”跨层级耦合链条,系统梳理了考虑算力弹性伸缩的智算中心与电网协同优化运行关键技术。将算力任务分为离线训练、在线推理和科学计算3类,分析了离线训练时间可平移与在线推理空间可迁移的差异化弹性特征,综述了物理能耗建模、时空负荷预测、异构算力池化调度、动态电压频率调整(DVFS)、断点续训和储能协同控制等关键技术,并总结了智算中心作为广义虚拟电厂参与需求响应与电力市场的联动机制;进一步基于深度强化学习将算电协同策略建模为马尔可夫决策过程,设计了涵盖“算随源动”时间维度策略与“算随电走”空间维度策略的云-边-端协同联动机制。研究表明:训练任务可通过任务重排、功率调节和断点续训等方式释放时间弹性,是削峰填谷的主力调节资源;推理任务可通过请求路由与跨区域迁移释放空间弹性,但受服务等级协议(SLA)约束呈低弹性特征;科学计算任务允许在特定时间窗口内错峰执行;设备调频与储能可补充快速功率响应,但整体调节能力受通信带宽、热安全、备用容量及设备折损等多重因素约束。Google碳感知算力调度、Microsoft电网友好型储能、冀北张家口电算协同虚拟电厂及宁夏中卫“源网荷储算”一体化等国内外工程实践已初步验证算力负荷参与新能源消纳、削峰填谷和辅助服务的技术可行性,但多数仍处于试点或示范阶段,规模化应用仍面临电算数据接口不统一、弹性边界难以可信量化、跨域安全约束耦合不足、多市场价值传导机制不完善及缺乏统一灵活性交易框架等瓶颈。未来应建立覆盖任务、设备、智算中心与电网的多时间尺度统一模型,在调度算法中引入硬件折损模型与算力中断代价以平衡电网需求响应红利与算力资产物理损伤,完善算力弹性评估、基线核验和电-算-碳协同机制,为“源网荷储算”融合运行提供支撑。
Intelligent computing centers, serving as high-power-density and highly fluctuating critical loads in new power systems, pose challenges to coordinated operation. To address these issues, key technologies for coordinated optimal operation of intelligent computing centers and power grids considering elastic scaling of computing power are systematically reviewed along the cross-layer coupling chain of "task—resource—energy consumption—power grid". Computing tasks are classified into three types: offline training, online inference, and scientific computing. The differentiated elasticity characteristics of temporal shiftability for offline training and spatial migratability in online inference are analyzed. Key technologies, including physical energy consumption modeling, spatiotemporal load prediction, heterogeneous computing resource pooling and scheduling, dynamic voltage and frequency scaling (DVFS), checkpoint-based training resumption, and coordinated energy storage control, are reviewed. In addition, the linkage mechanisms through which intelligent computing centers participate in demand response and electricity markets as generalized virtual power plants are summarized. Furthermore, computing-electricity coordination strategies are modeled as a Markov decision process based on deep reinforcement learning, and a cloud-edge-end coordinated linkage mechanism incorporating the temporal strategy of "computing follows energy sources" and the spatial strategy of "computing follows electricity" is designed. The results indicate that training tasks can release temporal elasticity through task reordering, power adjustment, and checkpoint-based training resumption, serving as the principal regulation resource for peak shaving and valley filling. Inference tasks can release spatial elasticity through request routing and cross-regional migration, but exhibit low elasticity due to service-level agreement (SLA) constraints. Scientific computing tasks can be executed during off-peak periods within specified time windows. Equipment frequency regulation and energy storage can provide supplementary rapid power response, although the overall regulation capability is constrained by multiple factors, such as communication bandwidth, thermal safety, reserve capacity, and equipment degradation. Engineering practices in China and abroad, including Google's carbon-aware computing scheduling, Microsoft's grid-friendly energy storage, the electricity-computing coordination virtual power plant in Zhangjiakou, northern Hebei, and the integrated "source-grid-load-storage-computing" project in Zhongwei, Ningxia, have preliminarily verified the technical feasibility of computing load participating in renewable energy consumption, peak shaving and valley filling, and ancillary services. However, most of these practices remain at the pilot or demonstration stage, and large-scale application still faces bottlenecks such as inconsistent electricity-computing data interfaces, difficulty in credibly quantifying elasticity boundaries, insufficient coupling of cross-domain security constraints, incomplete value transmission mechanisms across multiple markets, and the lack of a unified flexibility trading framework. Future efforts should establish a unified multi-timescale model covering tasks, equipment, intelligent computing centers, and power grids. Hardware degradation models and computing interruption costs should be incorporated into scheduling algorithms to balance the benefits of power grid demand response against physical damage to computing assets, and computing elasticity assessment, baseline verification, and electricity-computing-carbon coordinated mechanisms should be improved, thereby supporting integrated "source-grid-load-storage-computing" operation.
周孝信 , 赵强 , 张玉琼 , 等 . “双碳” 目标下我国能源电力系统发展趋势分析: 绿电替代与绿氢替代 [J]. 中国电机工程学报 , 2024 , 44 ( 17 ): 6707 - 6720 .
ZHOU Xiaoxin , ZHAO Qiang , ZHANG Yuqiong , et al . Analysis of the development trend of China's energy and power system under the dual carbon target:Green electricity substitution and green hydrogen substitution [J]. Proceedings of the CSEE , 2024 , 44 ( 17 ): 6707 - 6720 .
HAO Y , JIANG Y , LYU S . Can the digital economy effectively contribute to the clean energy transition:A provincial panel data analysis from China [J]. Environmental Innovation and Societal Transitions , 2025 , 56 : 101003 .
GAO D X , XIA N , LIU X Q , et al . Joint load adjustment and sleep management for virtualized gNBs in computing power networks [J]. IEEE Transactions on Wireless Communications , 2025 , 24 ( 3 ): 2067 - 2081 .
FAN Y B , LI Y J , NATALINO C , et al . Distributed model training task migration for hotspot management in intelligent computing center interconnection with tidal characteristics [J]. IEEE Transactions on Network and Service Management , 2025 , 22 ( 6 ): 5489 - 5502 .
AMIRALIZADEH S , DOYLEND J K . AI networking challenges:A system perspective [J]. IEEE Journal of Selected Topics in Quantum Electronics , 2025 , 31 ( 3 ): 1 - 7 .
许诺 , 毛聚 , 毛新述 , 等 . 算力部署、数据跨域流动与企业全要素生产率: 来自智算中心的证据 [J]. 中国工业经济 , 2025 ( 4 ): 61 - 79 .
XU Nuo , MAO Ju , MAO Xinshu , et al . Computility deployment, cross-domain data flow and enterprise total factor productivity: Evidence from intelligent computing centers [J]. China Industrial Economics , 2025 ( 4 ): 61 - 79 .
徐靖 , 林义芳 , 赵伟博 , 等 . 基于DCTimeXer的算力中心短期电力负荷预测方法 [J]. 煤炭经济研究 , 2026 , 46 ( 5 ): 117 - 123 .
XU Jing , LIN Yifang , ZHAO Weibo , et al . Short-term power load forecasting method for computing centers based on DCTimeXer [J]. Coal Economic Research , 2026 , 46 ( 5 ): 117 - 123 .
王继业 , 周春雷 , 李洋 , 等 . 数据中心关键技术和发展趋势研究综述 [J]. 电力信息与通信技术 , 2022 , 20 ( 8 ): 1 - 21 .
WANG Jiye , ZHOU Chunlei , LI Yang , et al . Review of key technologies and development trend of data center construction [J]. Electric Power Information and Communication Technology , 2022 , 20 ( 8 ): 1 - 21 .
LATIF I , NEWKIRK A C , CARBONE M R , et al . Single-node power demand during AI training: Measurements on an 8-GPU NVIDIA H100 system [J]. IEEE Access , 2025 , 13 : 61740 - 61747 .
RAMAKRISHNAN B , TURNER C , ALISSA H , et al . Understanding the impact of data center liquid cooling on energy and performance of machine learning and artificial intelligence workloads [J]. Journal of Electronic Packaging , 2025 , 147 ( 2 ): 021003 .
张慧峰 , 刘宁春 , 龙卫平 , 等 . 面向国产算力的超大规模智算集群网络: 关键挑战、技术途径与发展趋势 [J]. 电信科学 , 2025 , 41 ( 12 ): 1 - 26 .
ZHANG Huifeng , LIU Ningchun , LONG Weiping , et al . Hyperscale intelligent computing cluster networks for domestic computing power: Critical challenges, technical pathways, and future trends [J]. Telecommunications Science , 2025 , 41 ( 12 ): 1 - 26 .
陈锦前 , 郭少勇 , 刘畅 , 等 . 数据处理单元赋能的智算中心网络拥塞控制机制 [J]. 通信学报 , 2025 , 46 ( 2 ): 1 - 17 .
CHEN Jinqian , GUO Shaoyong , LIU Chang , et al . DPU empowered intelligent congestion control mechanism for the intelligent computing center network [J]. Journal on Communications , 2025 , 46 ( 2 ): 1 - 17 .
陈敏 , 郭庆来 , 井汤博 , 等 . 算电协同探索: HVDC侧储备一体电池灵活性挖掘 [J]. 电力建设 , 2025 , 46 ( 2 ): 1 - 12 .
CHEN Min , GUO Qinglai , JING Tangbo , et al . Exploring synergy of computing power and electricity: Unearthing flexibility of high-voltage direct current-side integrated backup and energy storage batteries [J]. Electric Power Construction , 2025 , 46 ( 2 ): 1 - 12 .
周师琦 , 王俊帆 , 赖俊升 , 等 . 结合贝叶斯Autoformer的多维自适应短期电力负荷概率预测方法 [J]. 电子与信息学报 , 2024 , 46 ( 12 ): 4432 - 4440 .
ZHOU Shiqi , WANG Junfan , LAI Junsheng , et al . Multi-view adaptive probabilistic load forecasting combining Bayesian autoformer network [J]. Journal of Electronics & Information Technology , 2024 , 46 ( 12 ): 4432 - 4440 .
张冬月 , 辛奇 , 韩博文 , 等 . 智算中心网络的亚毫秒级网络监控系统 [J]. 电信科学 , 2025 , 41 ( 8 ): 65 - 75 .
ZHANG Dongyue , XIN Qi , HAN Bowen , et al . Sub-millisecond-level network monitoring system for intelligent computing centers network [J]. Telecommunications Science , 2025 , 41 ( 8 ): 65 - 75 .
黄韬 , 周子翔 , 唐琴琴 , 等 . 面向自智算力网络的数字孪生: 架构与关键挑战 [J]. 通信学报 , 2025 , 46 ( 4 ): 255 - 271 .
HUANG Tao , ZHOU Zixiang , TANG Qinqin , et al . Digital twin for self-intelligent computing power networks: Architecture and key challenges [J]. Journal on Communications , 2025 , 46 ( 4 ): 255 - 271 .
慕国行 , 贺卫华 , 周自强 . 数字孪生驱动下微电网多智能体协调控制优化 [J]. 计算机仿真 , 2023 , 40 ( 12 ): 133 - 138, 149 .
MU Guoxing , HE Weihua , ZHOU Ziqiang . Optimization of multi-agent coordinated control of microgrid driven by digital twin [J]. Computer Simulation , 2023 , 40 ( 12 ): 133 - 138, 149 .
姚梦 . 新型电力系统下考虑储能调节策略与能量功率比的可靠性评估 [D]. 广州 : 华南理工大学 , 2024 .
YAO Meng . Reliability evaluation considering energy storage regulation strategy and energy-to-power ratio in new power system [D]. Guangzhou : South China University of Technology , 2024 .
HOGADE N , PASRICHA S . Game-theoretic deep reinforcement learning to minimize carbon emissions and energy costs for AI inference workloads in geo-distributed data centers [J]. IEEE Transactions on Sustainable Computing , 2024 , 10 ( 4 ): 628 - 641 .
王一航 , 张靖 , 何宇 , 等 . 计及风光不确定性的数据中心电算协同多目标优化策略 [J]. 电力工程技术 , 2026 , 45 ( 5 ): 40 - 49 .
WANG Yihang , ZHANG Jing , HE Yu , et al . A multi-objective power-computing collaborative optimization strategy for data centers considering wind and solar uncertainty [J]. Electric Power Engineering Technology , 2026 , 45 ( 5 ): 40 - 49 .
周钱雨凡 , 杨苹 , 万思洋 , 等 . 算力电力节点可调节资源的双向协同优化调度 [J]. 电力建设 , 2025 , 46 ( 2 ): 13 - 25 .
ZHOU Qianyufan , YANG Ping , WAN Siyang , et al . Bidirectional collaborative optimization scheduling of adjustable resources in computing node and power node [J]. Electric Power Construction , 2025 , 46 ( 2 ): 13 - 25 .
罗婧 . GPU集群中AI任务运行特征分析及调度优化 [D]. 武汉 : 武汉纺织大学 , 2023 .
LUO Jing . Characteristic analysis and scheduling optimization of AI workloads in GPU clusters [D]. Wuhan : Wuhan Textile University , 2023 .
连政行 . 面向智能电网需求侧响应的传输-计算联合优化策略研究 [D]. 秦皇岛 : 燕山大学 , 2024 .
LIAN Zhengxing . Research on transmission-computing joint optimization strategy for demand side response of smart grid [D]. Qinhuangdao : Yanshan University , 2024 .
URANBOLD T , LIMA L M . Energy solutions for data center: Comparative analysis of levelized cost of electricity (LCOE) and recent developments [J]. Energy Reports , 2025 , 14 : 3529 - 3535 .
郑芳芳 , 黄雨佳 , 孙秋野 , 等 . 基于激励惩罚平衡机制的算电一体化调度 [J/OL]. 电力系统自动化 , 1 - 10 ( 2026-02-11 )[ 2026-06-20 ]. https://link.cnki.net/urlid/32.1180.TP.20260210.1831.004 https://link.cnki.net/urlid/32.1180.TP.20260210.1831.004 .
ZHENG Fangfang , HUANG Yujia , SUN Qiuye , et al . Integrated dispatching of power calculation and electricity based on incentive and punishment balance mechanism [J/OL]. Automation of Electric Power Systems , 1 - 10 ( 2026-02-11 )[ 2026-06-20 ]. https://link.cnki.net/urlid/32.1180.TP.20260210.1831.004 https://link.cnki.net/urlid/32.1180.TP.20260210.1831.004 .
刘方 , 陈祥洲 , 周江昕 , 等 . 考虑数据中心共享储能与计算负荷时空迁移特性的虚拟电厂优化运行方法 [J]. 电网技术 , 2026 , 50 ( 5 ): 1848 - 1857 .
LIU Fang , CHEN Xiangzhou , ZHOU Jiangxin , et al . Optimal operation method for virtual power plant considering shared energy storage and spatiotemporal migration characteristics of computational load in data centers [J]. Power System Technology , 2026 , 50 ( 5 ): 1848 - 1857 .
周飞航 , 王灏 , 王海利 , 等 . 基于多智能体强化学习的电-碳-绿证耦合市场下多市场主体行为研究 [J]. 中国电力 , 2025 , 58 ( 4 ): 44 - 55 .
ZHOU Feihang , WANG Hao , WANG Haili , et al . Multi-entity behaviors in electricity-carbon-green certificate coupled markets based on multi-agent reinforcement learning [J]. Electric Power , 2025 , 58 ( 4 ): 44 - 55 .
TORRIJOS P , GÁMEZ J A , PUERTA J M . FedGES: A federated learning approach for Bayesian network structure learning [C]// Proceedings of Discovery Science . Springer , 2025 : 83 - 98 .
王志红 , 龚振炜 , 武丽丽 , 等 . 大模型时代我国高密度算力安全及发展研究 [J]. 中国工程科学 , 2026 , 28 ( 2 ): 72 - 84 .
WANG Zhihong , GONG Zhenwei , WU Lili , et al . Security and development of high-density computing power in China in the era of large language models [J]. Strategic Study of CAE , 2026 , 28 ( 2 ): 72 - 84 .
王永真 , 唐豪 , 魏一鸣 , 等 . 中国数据中心综合能耗及其灵活性预测 [J]. 北京理工大学学报(社会科学版) , 2025 , 27 ( 2 ): 12 - 18 .
WANG Yongzhen , TANG Hao , WEI Yiming , et al . Comprehensive energy consumption and flexibility forecast of data centers in China [J]. Journal of Beijing Institute of Technology(Social Sciences Edition) , 2025 , 27 ( 2 ): 12 - 18 .
王馨翊 , 王露 , 陈晓轩 , 等 . 数据中心高密度机柜的风冷系统及其散热能力分析 [J]. 制冷学报 , 2026 , 47 ( 1 ): 51 - 58 .
WANG Xinyi , WANG Lu , CHEN Xiaoxuan , et al . Analysis of air-cooling system and heat-dissipation capacity for high-density cabinets in data centers [J]. Journal of Refrigeration , 2026 , 47 ( 1 ): 51 - 58 .
朱宸 , 魏东黎 , 余海生 , 等 . 冷板式液冷技术应用于智算中心高密机房的方案分析 [J]. 电信工程技术与标准化 , 2024 , 37 ( S1 ): 174 - 179 .
ZHU Chen , WEI Dongli , YU Haisheng , et al . Analysis of the application of cold plate liquid cooling technology in high-density device room of intelligent data centers [J]. Telecom Engineering Technics and Standardization , 2024 , 37 ( S1 ): 174 - 179 .
0
浏览量
2
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