Method and Practice of Computing Power-Electric Power Coordination Based on Idle Computing Power Scheduling

FU Zhi; ZHENG Bowen; LIU Xiaochen; CHENG Wei; ZHANG Tao; LIU Xiaohua

doi:10.7500/AEPS20260205005

您当前的位置：

首页 >

文章列表页 >

Method and Practice of Computing Power-Electric Power Coordination Based on Idle Computing Power Scheduling

Computing Power-Electric Power Coordination | 更新时间：2026-09-01

- Method and Practice of Computing Power-Electric Power Coordination Based on Idle Computing Power Scheduling
- Automation of Electric Power Systems Vol. 50, Issue 17, Pages: 126-136(2026)
- 作者机构：
  
  1.清华大学建筑技术科学系，北京市 100084
  2.北京共绩科技有限公司，北京市 100080
  3.中国联通广东省分公司，广东省广州市 510627
- 作者简介：
- 基金信息：
  
  Ministry of Industry and Information Technology of the People’s Republic of China (cross-domain task orchestration system for heterogeneous computing power), Beijing Municipal Natural Science Foundation of China;Youth Talent Support Program of China Association for Science and Technology(QY25032);Young Elite Scientists Sponsorship Program by CAST(2023QNRC001)
- DOI：10.7500/AEPS20260205005
  CLC：
- Received：05 February 2026，
  
  Revised：2026-04-29，
  
  Online First：20 July 2026，
  
  Published：10 September 2026
- 稿件说明：
移动端阅览
付智,郑博文,刘效辰等.基于闲时算力调度的算电协同方法与实践[J].电力系统自动化,2026,50(17):126-136.

FU Zhi,ZHENG Bowen,LIU Xiaochen,et al.Method and Practice of Computing Power-Electric Power Coordination Based on Idle Computing Power Scheduling[J].Automation of Electric Power Systems,2026,50(17):126-136.
付智,郑博文,刘效辰等.基于闲时算力调度的算电协同方法与实践[J].电力系统自动化,2026,50(17):126-136. DOI： 10.7500/AEPS20260205005.

FU Zhi,ZHENG Bowen,LIU Xiaochen,et al.Method and Practice of Computing Power-Electric Power Coordination Based on Idle Computing Power Scheduling[J].Automation of Electric Power Systems,2026,50(17):126-136. DOI： 10.7500/AEPS20260205005.

摘要

数据中心能耗持续增长对电力系统构成严峻挑战，尤其是大语言模型推理计算任务的实时响应需求导致了显著的用电尖峰负荷。闲时算力资源规模庞大且具备时空调节灵活性，是实现算电协同的重要潜力资源。为充分发挥闲时算力的柔性调节能力，提出了一种基于闲时算力调度的算电协同方法，以实现计算任务的高稳定性和低成本。首先，分析了图形处理器单机“性能-功耗”非线性特性；其次，构建了面向闲时算力的“时域-地域”调度策略；最后，在真实闲时算力调度平台上开展了工程实践验证。实验结果表明，时域调节使算力系统总功耗降低17%以上，地域调度使计算成本降低近18%，地区绿电利用率提高近20%。所提方法在降低计算能耗、促进可再生能源利用等方面具有显著效果和推广潜力，为新型电力系统和算力网络的协同低碳发展提供了可行方案。

Abstract

The continuous growth of energy consumption in data centers poses severe challenges to power systems. In particular

the real-time response requirements for large language model inference tasks generate significant electricity peak loads. Idle computing resources feature a large scale and spatio-temporal regulation flexibility

serving as high-potential resources for realizing computing power-electric power coordination. To fully exploit the flexibility regulation capability of idle computing power

this paper proposes a computing power-electric power coordination method based on idle computing power scheduling to achieve high stability and low cost for computing tasks. Firstly

the nonlinear “performance-power consumption” characteristics of a single graphics processing unit are analyzed. Secondly

a temporal-regional scheduling strategy oriented to idle computing power is constructed. Finally

engineering practical verification is conducted on a real idle computing power scheduling platform. The experimental results show that the temporal regulation reduces the total power consumption of the computing power system by more than 17%

the regional scheduling cuts computing costs by nearly 18%

and improves the regional green power utilization rate by approximately 20%. The proposed method achieves remarkable performance and possesses broad promotion potential in reducing computing energy consumption and boosting renewable energy utilization

which provides a feasible solution for the coordinated low-carbon development of new power systems and computing power networks.

关键词

Keywords

references

International Energy Agency （IEA） . Energy and AI ［R］. Paris ： International Energy Agency ， 2025 .

中国信息通信研究院产业与规划所，内蒙古和林格尔新区管委会 . 绿色算力发展研究报告（2025年）［R］. 北京：中国信息通信研究院， 2025 .

China Academy of Information and Communications Technology ， Institute of Industry and Planning ， and Administrative Committee of Horinger New Area ， Inner Mongolia . Green computing power development research report （2025）［R］. Beijing ： China Academy of Information and Communications Technology ， 2025 .

Chen S R ， Li P ， Ji H R ， et al . Operational flexibility of active distribution networks with the potential from data centers ［J］. Applied Energy ， 2021 ， 293 ： 116935 .

Alaperä I ， Honkapuro S ， Paananen J . Data centers as a source of dynamic flexibility in smart girds ［J］. Applied Energy ， 2018 ， 229 ： 69 - 79 .

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 .

丁巧宜，王梓耀，潘振宁，等 . 面向电量-调频-容量市场的数据中心园区算力及电力资源规划［J］. 电力系统自动化， 2024 ， 48 （ 1 ）： 59 - 66 .

Ding Qiaoyi ， Wang Ziyao ， Pan Zhenning ， et al . Planning of computing power and electric power resources in data center parks for electricity， frequency regulation and capacity markets ［J］. Automation of Electric Power Systems ， 2024 ， 48 （ 1 ）： 59 - 66 .

数据中心绿色低碳发展专项行动计划［EB/OL］.［ 2025-07-03 ］. https：//www.gov.cn/zhengce/zhengceku/202407/content_6964 165.htm https://www.gov.cn/zhengce/zhengceku/202407/content_6964165.htm .

Special action plan for green and low-carbon development of data centers ［EB/OL］. ［ 2025-07-03 ］. https：//www.gov.cn/zhengce/zhengceku/202407/content_6964165.htm https://www.gov.cn/zhengce/zhengceku/202407/content_6964165.htm .

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 .

绿色算力产业发展工作成效显著［EB/OL］.［ 2025-01-19 ］. https：//app.people.cn/h5/detail/normal/6152684713772032 https://app.people.cn/h5/detail/normal/6152684713772032 .

Remarkable achievements in the development of the green computing power industry ［EB/OL］. ［ 2025-01-19 ］. https：//app.people.cn/h5/detail/normal/6152684713772032 https://app.people.cn/h5/detail/normal/6152684713772032 .

Morgan P . The battle begins for AI inference compute in the datacenter ［EB/OL］. ［ 2025-09-10 ］. https：//www.nextplatf orm.com/2024/09/10/the-battle-begins-for-ai-inference-compu te-in-the-datacenter/#：~：text=applications%20or%20to%20 create%20entirely https://www.nextplatform.com/2024/09/10/the-battle-begins-for-ai-inference-compute-in-the-datacenter/#:~:text=applications%20or%20to%20create%20entirely ，ging%20needsprocesses%20with%20modest%20packa.

Desislavov R ， Martínez-Plumed F ， Hernández-Orallo J . Trends in AI inference energy consumption： beyond the performance-vs-parameter laws of deep learning ［J］. Sustainable Computing： Informatics and Systems ， 2023 ， 38 ： 100857 .

胡明，李晓华，蔡泽祥，等 . 基于计算负荷预测的配用电终端算力供需动态均衡方法［J］. 电力自动化设备， 2024 ， 44 （ 12 ）： 52 - 60 .

Hu Ming ， Li Xiaohua ， Cai Zexiang ， et al . Dynamic balancing method for supply and demand of power distribution and consumption terminal arithmetic based on computational load forecasting ［J］. Electric Power Automation Equipment ， 2024 ， 44 （ 12 ）： 52 - 60 .

秦雪雪，李泽宁，薛屹洵，等 . 计及建筑热动态特性的数据中心时空协调两阶段鲁棒优化方法［J］. 电力系统自动化， 2025 ， 49 （ 11 ）： 59 - 69 .

Qin Xuexue ， Li Zening ， Xue Yixun ， et al . Two-stage robust optimization method for spatio-temporal coordination of data center considering thermal dynamic characteristics of buildings ［J］. Automation of Electric Power Systems ， 2025 ， 49 （ 11 ）： 59 - 69 .

Cao Y J ， Cao F ， Wang Y J ， et al . Managing data center cluster as non-wire alternative： a case in balancing market ［J］. Applied Energy ， 2024 ， 360 ： 122769 .

Yu L ， Jiang T ， Zou Y L . Distributed real-time energy management in data center microgrids ［J］. IEEE Transactions on Smart Grid ， 2018 ， 9 （ 4 ）： 3748 - 3762 .

Zhang L T ， Deng J K ， Li Q F ， et al . Research on data center computing resources and energy load co-optimization considering spatial-temporal allocation ［J］. Computers and Electrical Engineering ， 2024 ， 116 ： 109206 .

Luo J Y ， Rao L ， Liu X . Spatio-temporal load balancing for energy cost optimization in distributed Internet data centers ［J］. IEEE Transactions on Cloud Computing ， 2015 ， 3 （ 3 ）： 387 - 397 .

Ilager S ， Muralidhar R ， Rammohanrao K ， et al . A data-driven frequency scaling approach for deadline-aware energy efficient scheduling on graphics processing units （GPUs）［C］// 20th IEEE/ACM International Symposium on Cluster ， Cloud and Internet Computing ， 2020 .

Kakolyris A K ， Masouros D ， Xydis S ， et al . SLO-aware GPU DVFS for energy-efficient LLM inference serving ［J］. IEEE Computer Architecture Letters ， 2024 ， 23 （ 2 ）： 150 - 153 .

Maliakel P J ， Ilager S ， Brandic I . Characterizing LLM inference energy-performance tradeoffs across workloads and GPU scaling ［PP/OL］. ［ 2025-08-05 ］. https：//arxiv.org/abs/2501.08219 https://arxiv.org/abs/2501.08219 .

Wang Y ， Guo Q L ， Chen M . Providing load flexibility by reshaping power profiles of large language model workloads ［J］. Advances in Applied Energy ， 2025 ， 19 ： 100232 .

Sun J J ， Fu Z ， Ma J H ， et al . Sharing computing power： enhancing energy efficiency and reducing carbon emissions ［J］. IEEE Energy Sustainability Magazine ， 2026 ， 2 （ 1 ）： 103 - 119 .

中国信息通信研究院 . 云计算蓝皮书（2025年）［R］. 北京：中国信息通信研究院， 2025 .

China Academy of Information and Communications Technology . Cloud computing blue book （2025）［R］. Beijing ： China Academy of Information and Communications Technology ， 2025 .

都芃 . 利用率仅三成，“沉睡”的算力如何激活［EB/OL］.［ 2025-07-17 ］. https：//gdstc.gd.gov.cn/kjzx_n/mtjj/content/post_474 5744.html https://gdstc.gd.gov.cn/kjzx_n/mtjj/content/post_4745744.html .

Du Peng . With a utilization rate of only 30 %， how can we awaken “dormant” computing power［EB/OL］. ［ 2025-07-17 ］. https：//gdstc.gd.gov.cn/kjzx_n/mtjj/content/post_4745744.html https://gdstc.gd.gov.cn/kjzx_n/mtjj/content/post_4745744.html .

全国数据资源统计调查工作组 . 全国数据资源调查报告（2024年）［R］. 北京：全国数据资源统计调查工作组， 2024 .

National Data Resources Statistical Survey Working Group . National data resources survey report （2024）［R］. Beijing ： National Data Resources Statistical Survey Working Group ， 2024 .

Steam硬件和软件调查［EB/OL］.［ 2025-10-05 ］. https：//store.steampowered.com/hwsurvey/Steam-Hardware-Software-Surv ey-Welcome-to-Steam https://store.steampowered.com/hwsurvey/Steam-Hardware-Software-Survey-Welcome-to-Steam .

Steam hardware and software survey ［EB/OL］. ［ 2025-10-05 ］. https：//store.steampowered.com/hwsurvey/Steam-Hardware-Software-Survey-Welcome-to-Steam https://store.steampowered.com/hwsurvey/Steam-Hardware-Software-Survey-Welcome-to-Steam .

Tang Z Q ， Zhang F M ， Zhou X J ， et al . Pricing model for dynamic resource overbooking in edge computing ［J］. IEEE Transactions on Cloud Computing ， 2023 ， 11 （ 2 ）： 1970 - 1984 .

Sabyasachi A S ， Kabir H M D ， Abdelmoniem A M ， et al . A resilient auction framework for deadline-aware jobs in cloud spot market ［C］// IEEE 36th Symposium on Reliable Distributed Systems ， 2017 .

Al-Theiabat H ， Al-Ayyoub M ， Alsmirat M ， et al . A deep learning approach for Amazon EC2 spot price prediction ［C］// IEEE/ACS 15th International Conference on Computer Systems and Applications （AICCSA）， 2018 .

Baughman M ， Haas C ， Wolski R ， et al . Predicting Amazon spot prices with LSTM networks ［C］// 9th Workshop on Scientific Cloud Computing ， 2018 .

Li J C ， Wan Q ， Yu Z P . Research on network security algorithms in distributed computing environment ［C］// SPIE Conference on International Conference on Internet of Things and Machine Learning （IoTML 2023）， 2023 .

杨明烜，洪学海，唐宏伟 . 基于任务资源需求预测的人工智能算力调度［J］. 高技术通讯， 2024 ， 34 （ 5 ）： 475 - 485 .

Yang Mingxuan ， Hong Xuehai ， Tang Hongwei . Artificial intelligence computing power cluster scheduling based on task resource demand prediction ［J］. Chinese High Technology Letters ， 2024 ， 34 （ 5 ）： 475 - 485 .

Guilloteau Q ， Richard O ， Robu B ， et al . Controlling the injection of best-effort tasks to harvest idle computing grid resources ［C］// 25th International Conference on System Theory， Control and Computing （ICSTCC）， 2021 .

Duan X ， Ye K J . Leftover： improving large language model inference efficiency by leveraging idle resources ［C］// International Conference on High Performance Big Data and Intelligent Systems （HDIS）， 2023 .

Sun J J ， Yan Y J ， Wang Z Y ， et al . Privacy-preserving coordinated operation of cross-enterprise data centers ［J］. Applied Energy ， 2025 ， 383 ： 125343 .

共绩算力 . 算了么闲时算力接入平台［EB/OL］.［ 2025-10-08 ］. https：//suanleme.cn https://suanleme.cn .

Comnergy . Suanleme idle computing power accesses platform ［EB/OL］. ［ 2025-10-08 ］. https：//suanleme.cn https://suanleme.cn .

共绩算力 . 闲时算力调度平台［EB/OL］.［ 2025-10-08 ］. https：//suanli.cn https://suanli.cn .

Comnergy . Idle computing power scheduling platform ［EB/OL］. ［ 2025-10-08 ］. https：//suanli.cn https://suanli.cn .

可思为，董萍，马铭宇，等 . 考虑风光荷时空互补的多能源绿色数据中心多目标配置方法［J］. 电力系统保护与控制， 2024 ， 52 （ 22 ）： 22 - 33 .

Ke Siwei ， Dong Ping ， Ma Mingyu ， et al . A multi-objective allocation method for multi-energy green data centers considering wind， solar and load spatial-temporal complementarity ［J］. Power System Protection and Control ， 2024 ， 52 （ 22 ）： 22 - 33 .

零克云 . GPU算力租赁［EB/OL］.［ 2025-10-08 ］. https：//gpulink.cc/computility-market https://gpulink.cc/computility-market .

ZeroKe Cloud . GPU computing power rental ［EB/OL］. ［ 2025-10-08 ］. https：//gpulink.cc/computility-market https://gpulink.cc/computility-market .

刘祎泽，向月 . 计及负荷转移需求响应的低碳数据中心光储容量优化配置［J］. 电力自动化设备， 2024 ， 44 （ 7 ）： 149 - 155

Liu Yize ， Xiang Yue . Optimal allocation of photovoltaic and storage capacity in low carbon data center considering load transfer demand response ［J］. Electric Power Automation Equipment ， 2024 ， 44 （ 7 ）： 149 - 155 .

国网青海省电力公司 . 关于 2025 年9月代理购电价格的公告［EB/OL］.［ 2025-08-05 ］. https：//csc-static.sgcc.com.cn：28888/omg-static/99308281750147514938600207141603.html https://csc-static.sgcc.com.cn:28888/omg-static/99308281750147514938600207141603.html .

State Grid Qinghai Electric Power Company . Announcement on the agent power purchasing price for September 2025 ［EB/OL］. ［ 2025-08-05 ］. https：//csc-static.sgcc.com.cn：28888/omg-static/99308281750147514938600207141603.html https://csc-static.sgcc.com.cn:28888/omg-static/99308281750147514938600207141603.html .

Holtzman A ， Buys J ， Du L ， et al . The curious case of neural text degeneration ［PP/OL］. ［ 2025-08-05 ］. https：//arxiv.org/abs/1904.09751 https://arxiv.org/abs/1904.09751 .

Wilhelm P ， Wittkopp T ， Kao O . Beyond test-time compute strategies： advocating energy-per-token in LLM inference ［C］// 5th Workshop on Machine Learning and Systems ， 2025 .

Xiao C J ， Cai J ， Zhao W L ， et al . Densing law of LLMs ［J］. Nature Machine Intelligence ， 2025 ， 7 （ 11 ）： 1823 - 1833 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Research Status and Prospects of Modeling and Applications for Computing Power-Electric Power Coordination

Load Flexibility of Intelligent Computing Centers Aimed for Computing Power-Electric Power Coordination: Challenges and Prospects

Optimization Method for Coordinated Operation of Computing Power Network and Electric Power Network Under Multi-source Uncertainty Risk Constraints

Coordinated Scheduling of Computing Power and Electric Power in Data Centers Considering Adaptive Adjustment of Computing Resources

Coordinated Optimal Scheduling of Computing Power-Electric Power Resources in Multi-data Centers Based on Accelerated Deep Reinforcement Learning

Related Author

YANG Meng

WENG Xi

SHEN Chen

CHEN Yue

WANG Yi

GUO Qinglai

MA Ding

YE Yujian

Related Institution

Department of Mechanical and Automation Engineering, The Chinese University of Hong Kong

Guanghua School of Management, Peking University

Department of Electrical Engineering, Tsinghua University

School of Electrical Engineering, Southeast University

Center of AI for Science and Engineering, Shenzhen Loop Area Institute

⁰