新型电力系统下算电协同发展关键技术及趋势

doi:10.12267/j.issn.2096-5931.2026.07.001

信息通信技术与政策 ›› 2026, Vol. 52 ›› Issue (7): 2-10.doi: 10.12267/j.issn.2096-5931.2026.07.001

新型电力系统下算电协同发展关键技术及趋势

Key technologies and trends for computing-power synergistic development under the new power system

石鑫¹^,², 王雁河², 高峰², 张頔³

¹ 内蒙古工业大学新能源学院, 鄂尔多斯 017010
² 清华大学能源互联网创新研究院, 北京 100085
³ 国网北京海淀供电公司, 北京 100102

收稿日期:2026-06-15 出版日期:2026-07-25 发布日期:2026-07-30
作者简介:
石鑫, 内蒙古工业大学新能源学院副研究员,清华大学能源互联网创新研究院副研究员,长期从事新型能源电力系统建模分析、人工智能及应用等方面的研究工作
王雁河, 清华大学能源互联网创新研究院副研究员,长期从事能源互联网、交能融合等方面的研究工作
高峰, 清华大学能源互联网创新研究院教授级高级工程师,副院长,长期从事能源互联网与能源行业数字化转型等方面的研究工作
张頔, 国网北京海淀供电公司高级工程师,长期从事大数据分析、区块链、人工智能及应用等方面的研究工作

SHI Xin¹^,², WANG Yanhe², GAO Feng², ZHANG Di³

¹ School of New Energy, Inner Mongolia University of Technology, Ordos 017010, China
² Energy Internet Research Institute, Tsinghua University, Beijing 100085, China
³ State Grid Beijing Haidian Power Supply Company, Beijing 100102, China

Received:2026-06-15 Online:2026-07-25 Published:2026-07-30

摘要/Abstract

摘要：

在新型电力系统建设和新一代人工智能(Artificial Intelligence,AI)快速发展的背景下,算力需求持续增长,对电力供应的稳定性、可靠性、绿电占比等提出了更高要求,与以风光为主体的新型电力系统形成了结构性矛盾,电算协同发展势在必行。围绕算电协同的内涵、发展现状、关键技术及面临挑战展开研究,首先阐释了算电协同发展的基本概念及内涵,并对其发展现状进行梳理;其次从算力负荷预测、高可靠供电、绿电直供、跨时空调度、市场交易5个方面系统总结分析了算电协同发展关键技术路径;最后对算电协同面临的挑战及未来发展趋势进行探讨分析。研究旨在为算电协同发展提供一定的理论和技术支持,为相关从业者提供参考。

关键词: 新型电力系统, 新一代AI, 算电协同, 跨时空调度

Abstract:

Under the new power system construction and the rapid development of the new generation of Artificial Intelligence (AI), the continuous growth of computing demand has placed high requirements on the stability, reliability and green electricity ratio, creating a structural contradiction with the new power system dominated by wind and solar energy, making computing-power synergistic development imperative. This paper studies the connotation, development situation, key technologies and challenges of computing-power synergy. It first expounds the basic concepts and connotation of computing-power synergistic development and reviews its development situation. Then, it systematically examines key technical paths from five aspects, namely, computing load forecasting, highly reliable power supply, green power direct supply, spatiotemporal scheduling and market trading. Finally, it discusses the challenges and future development trends of computing-power synergy. The research results aim to provide a certain theoretical basis and application guidance for the development of computing-power synergy, and serve as a reference for relevant practitioners.

Key words: new power system, new-generation AI, computing-power synergy, spatiotemporal scheduling

中图分类号:

石鑫, 王雁河, 高峰, 张頔. 新型电力系统下算电协同发展关键技术及趋势[J]. 信息通信技术与政策, 2026, 52(7): 2-10.

SHI Xin, WANG Yanhe, GAO Feng, ZHANG Di. Key technologies and trends for computing-power synergistic development under the new power system[J]. Information and Communications Technology and Policy, 2026, 52(7): 2-10.

导出引用管理器 EndNote|Ris|BibTeX

链接本文:

http://ictp.caict.ac.cn/CN/10.12267/j.issn.2096-5931.2026.07.001

http://ictp.caict.ac.cn/CN/Y2026/V52/I7/2

图/表 6

图1 典型延迟敏感型与容忍型负荷曲线

图2 基于新一代AI的算力负荷预测方法图谱

图3 传统交流供电与SST柔直供电架构对比

图4 “荷随源转”和“源随荷建”模式对比

图5 算力负荷跨时空调度

图6 算力中心参与电力市场

参考文献 17

[1]	邵子豪, 王志浩, 周晓茂, 等. 算电协同研究综述:架构、关键技术与展望[J]. 通信学报, 2025, 46(10):287-308.
[2]	李洁, 王月, 姜凌菲. 算力电力协同发展内涵与关键举措研究[J]. 信息通信技术与政策, 2025, 51(2):2-8. doi: 10.12267/j.issn.2096-5931.2025.02.001
[3]	Radovanovic A, Koningstein R, Schneider I, et al. Carbon-aware computing for datacenters[J]. IEEE Transactions on Power Systems, 2021, 38(2):1270-1280. doi: 10.1109/TPWRS.2022.3173250 URL
[4]	Kez D A, Foley A M, Morrow D J. Analysis of fast frequency response allocations in power systems with high system non-synchronous penetrations[J]. IEEE Transactions on Industry Applications, 2022, 58(3): 3087-3101. doi: 10.1109/TIA.2022.3160997 URL
[5]	Melnatami C C. AI data centers and U.S. grids: a 2026-2030 regional forecast[C]// 2026 IEEE Green Technologies Conference (GreenTech). Boulder, USA, 2026: 1-6.
[6]	Colangelo P, Coskun A K, Megrue J, et al. AI data centers as grid-interactive assets[J]. Nature Energy, 2026(11):254-261.
[7]	Zhang D, Chen K, Li W, et al. A load forecasting method for data centers based on MHSA-GRU[C]// 2026 11th Asia Conference on Power and Electrical Engineering (ACPEE). Macao, 2026:931-936.
[8]	Mughees M, Li Y, Chen Y, et al. Short-term load forecasting for AI-data center[PP/OL]. arXiv(2025-03-10)[2026-06-14].https//arxiv.org/abs/2503.07756.
[9]	Wang H, Chao J. ENNP: an enhanced neural networks-based power consumption prediction algorithm to deep learning-based workloads on AI-enabled data centers[J]. Cluster Computing, 2026, 29: 36. doi: 10.1007/s10586-025-05840-w
[10]	Tripathi S, Priyadarshni P, Misra R, et al. Multilayer multivariate forecasting network for precise resource utilization prediction in edge data centers[J]. Future Generation Computer Systems, 2025, 166:107692. doi: 10.1016/j.future.2024.107692 URL
[11]	Liu X. Neural network-based prediction method for data center resource utilization[C]// 2024 International Seminar on Artificial Intelligence, Computer Technology and Control Engineering (ACTCE). IEEE, 2024: 452-456.
[12]	Xia S, Yang J, Ruan X, et al. ALAS: attention mechanism-based and load-aware active scheduling for server resource optimization in complex scenarios[J]. Displays, 2026, 93:103438. doi: 10.1016/j.displa.2026.103438 URL
[13]	Jiang H, Qu B, Zhu J, et al. HyperLoad: a cross-modality enhanced large language model-based framework for green data center cooling load prediction[C]//Association for the Advancement of Artificial Intelligence.Proceedings of the AAAI Conference on Artificial Intelligence. Washington, DC, USA: AAAI Press, 2026:480-488.
[14]	Huang S, Dong J, Peng Z, et al. Resource load prediction of cloud computing based on STL-DeepAR-HW composite model[J]. Computer Applications and Software, 2025, 42(8): 367-373.
[15]	GB/T 45418—2025 配电网通用技术导则[S].
[16]	中国信息通信研究院. 算力电力协同发展研究报告(2025年)[R], 2025.
[17]	郑颖. 这份《意见》带出盎然春色[J]. 中国石油石化, 2025(8): 44-45.

新型电力系统下算电协同发展关键技术及趋势

Key technologies and trends for computing-power synergistic development under the new power system

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 6

参考文献 17

相关文章 10

编辑推荐

Metrics

本文评价

[1]	何琪, 陈翔, 江鹏, 赵子岩. 算电协同标准体系规划研究[J]. 信息通信技术与政策, 2026, 52(7): 11-15.
[2]	禤亮, 张子昊, 徐欢, 赵亮. 电碳算融合视角下电网企业算力Token化商业模式与调度机制研究[J]. 信息通信技术与政策, 2026, 52(7): 16-21.
[3]	严茂胜, 唐诗琦, 李腾达. 人工智能赋能算电协同的关键技术与实践探索[J]. 信息通信技术与政策, 2026, 52(7): 22-30.
[4]	张晓丹. 算电协同产业发展面临的现状与路径思考[J]. 信息通信技术与政策, 2026, 52(7): 31-37.
[5]	毛宇博, 孙小强, 沈岸平, 许伟, 陈龙泉. 面向算电协同的算力基础设施能效评估体系与标准化演进研究[J]. 信息通信技术与政策, 2026, 52(7): 38-46.
[6]	王超. 算电协同典型示范区实践模式比较与优化路径研究:以乌兰察布与庆阳为例[J]. 信息通信技术与政策, 2026, 52(7): 47-54.
[7]	杨思维, 赵亚楠, 刘宇翔. 算电协同的战略升维与制度破局:从结构性错配到机制重构[J]. 信息通信技术与政策, 2026, 52(7): 55-61.
[8]	蒋贱, 郝建平, 唐文佳, 何清素, 田继明, 葛乐矣, 王雪彩. “算电协同”方向下的中国算力经济发展逻辑与优化路径[J]. 信息通信技术与政策, 2026, 52(7): 62-71.
[9]	贾宝付, 高立山. 算电协同政策落地与产业生态优化路径研究[J]. 信息通信技术与政策, 2026, 52(7): 72-78.
[10]	冯宝, 赵子岩, 卞宇翔, 吕超, 王开创. 面向新型电力系统的量子保密通信技术及应用[J]. 信息通信技术与政策, 2025, 51(7): 33-42.