Neural Branching Power-Computation Network Fast Optimization Method Considering Computing Demand Response

doi:10.16183/j.cnki.jsjtu.2023.616

Abstract

Abstract:

The rapid development of data centers makes it possible for them to participate in power system dispatching as demand response. By dispatching the computing resources in data centers between regions, it is possible to achieve energy saving, emission reduction, and cost saving. However, considering the demand response of data center computing resources in power system dispatching faces the problem of insufficient computing speed. To address this issue, a neural branching power-computation network fast optimization method considering computing demand response is proposed. First, a unit commitment double-layer model considering computing power resource demand response is established. Then, the graph convolutional neural network and the branch and bound method are combined and applied to the double-layer model. Through historical data training, the neural branching power-computation network fast optimization method considering computing demand response has the ability to quickly determine the order of branch and bound variables and minimize the number of iterations, which significantly improves the solution speed and realizes the fast solution of unit combination demand response considering data center computing resources. The performance of the proposed method is verified in the simulation scenario of “east data, west computing” project. The proposed method achieves an average solution time reduction of 39.1% compared to the pseudo-cost branching algorithm, 38.1% compared to the commercial solver CPLEX, and 13.5% compared to the machine learning-based optimization acceleration algorithm Extratrees, the average solution time is reduced by 13.5%. In addition, the system coordinated dispatching frequency is increased from 1 time/h to 4 times/h, and the maximum potential reduction in wind curtailment of the total wind power generation during 24 h accounts for 17.42%.

Key words: mixed integer linear programming (MILP), unit commitment, demand response, Internet data center (IDC), neural branching

CLC Number:

TM732

ZHANG Lei, LI Ran, TANG Lun, CHEN Sijie, ZHAO Shizhen, SU Fu. Neural Branching Power-Computation Network Fast Optimization Method Considering Computing Demand Response[J]. Journal of Shanghai Jiao Tong University, 2025, 59(11): 1592-1602.

Figures/Tables 11

Fig.1

Fig.2

Fig.3

Fig.4

Fig.5

Tab.1

Fig.6

Tab.2

Tab.3

Tab.4

Fig.7

References 24

[1]	LI Y, TIAN X, HU J, et al.Research on reactive power planning technology of power grid containing UHVDC system with high proportion of renewable energy integration[C]//2020 IEEE International Conference on High Voltage Engineering and Application (ICHVE). Beijing, China: IEEE, 2020: 1-4.
[2]	慈松, 刘前卫, 康重庆, 等. 从“信息-能量”基本关系看信息能源深度融合[J]. 中国电机工程学报, 2021, 41(7): 2289-2297.
	CI Song, LIU Qianwei, KANG Chongqing, et al. Fundamental exploration into ICT-energy fusion[J]. Proceedings of the CSEE, 2021, 41(7): 2289-2297.
[3]	曹雨洁, 丁肇豪, 王鹏, 等. 能源互联网背景下数据中心与电力系统协同优化(二): 机遇与挑战[J]. 中国电机工程学报, 2022, 42(10): 3512-3527.
	CAO Yujie, DING Zhaohao, WANG Peng, et al. Coordinated operation for data center and power system in the context of energy Internet (II): Opportunities and challenges[J]. Proceedings of the CSEE, 2022, 42(10): 3512-3527.
[4]	杨挺, 姜含, 侯昱丞, 等. 基于计算负荷时-空双维迁移的互联多数据中心碳中和调控方法研究[J]. 中国电机工程学报, 2022, 42(1): 164-177.
	YANG Ting, JIANG Han, HOU Yucheng, et al. Study on carbon neutrality regulation method of interconnected multi-datacenter based on spatio-temporal dual-dimensional computing load migration[J]. Proceedings of the CSEE, 2022, 42(1): 164-177.
[5]	丁肇豪, 曹雨洁, 张素芳, 等. 能源互联网背景下数据中心与电力系统协同优化(一): 数据中心能耗模型[J]. 中国电机工程学报, 2022, 42(9): 3161-3177.
	DING Zhaohao, CAO Yujie, ZHANG Sufang, et al. Coordinated operation for data center and power system in the context of energy Internet (I): Energy demand management model of data center[J]. Proceedings of the CSEE, 2022, 42(9): 3161-3177.
[6]	YU L, JIANG T, ZOU Y. Price-sensitivity aware load balancing for geographically distributed internet data centers in smart grid environment[J]. IEEE Transactions on Cloud Computing, 2018, 6(4): 1125-1135. doi: 10.1109/TCC.2016.2564406 URL
[7]	曹晓峻, 高赐威, 李德智, 等. 数据网络与电力网络混合运行建模及其参与系统经济运行[J]. 中国电机工程学报, 2018, 38(5): 1448-1456.
	CAO Xiaojun, GAO Ciwei, LI Dezhi, et al. Mixed operation model of data network and power network and its participation in the economic operation of power system[J]. Proceedings of the CSEE, 2018, 38(5): 1448-1456.
[8]	陈敏, 高赐威, 郭庆来, 等. 互联网数据中心负荷时空可转移特性建模与协同优化: 驱动力与研究架构[J]. 中国电机工程学报, 2022, 42(19): 6945-6958.
	CHEN Min, GAO Ciwei, GUO Qinglai, et al. Modeling and coordinated optimization for spatiotemporal load regulation potentials of internet data centers: Motivation and architecture[J]. Proceedings of the CSEE, 2022, 42(19): 6945-6958.
[9]	刘晶. 东数西算大棋局: 盘活西部数据中心是关键[N]. 中国电子报, 2022-04-26(005).
	LIU Jing. East number west calculate big chess game: Revitalize the western data center is the key[N]. China Electronic News, 2022-04-26(005).
[10]	吕天文. “双碳”目标下 “东数西算”节能新路径[J]. 通信世界, 2022(6): 26-28.
	LÜ Tianwen. A new energy-saving path of “east data, west computing” under the “double carbon” goal[J]. Communication World, 2022(6): 26-28.
[11]	ZHENG J, CHIEN A, SUH S. Mitigating curtailment and carbon emissions through load migration between data centers[J]. Joule, 2020, 4(10): 2208-2222. doi: 10.1016/j.joule.2020.08.001 URL
[12]	周炳海, 宗师. 冷链物流越库调度的拉格朗日松弛算法[J]. 控制理论与应用, 2020, 37(3): 505-512.
	ZHOU Binghai, ZONG Shi. Lagrangian relaxation algorithm for cross-dock scheduling & problems in cold-chain logistics[J]. Control Theory & Applications, 2020, 37(3): 505-512.
[13]	赵芮, 顾幸生. 求解零空闲流水车间调度问题的离散正弦优化算法[J]. 上海交通大学学报, 2020, 54(12): 1291-1299. doi: 10.16183/j.cnki.jsjtu.2019.321
	ZHAO Rui, GU Xingsheng. A discrete sine optimization algorithm for no-idle flow-shop scheduling problem[J]. Journal of Shanghai Jiao Tong University, 2020, 54(12): 1291-1299.
[14]	潘险险, 陈霆威, 许志恒, 等. 适应多场景的微电网一体化柔性规划方法[J]. 上海交通大学学报, 2022, 56(12): 1598-1607. doi: 10.16183/j.cnki.jsjtu.2021.402
	PAN Xianxian, CHEN Tingwei, XU Zhiheng, et al. A multi-scenario integrated flexible planning method for microgrid[J]. Journal of Shanghai Jiao Tong University, 2022, 56(12): 1598-1607.
[15]	蒋小康, 张朋, 吕佑龙, 等. 基于混合蚁群算法的半导体生产线炉管区调度方法[J]. 上海交通大学学报, 2020, 54(8): 792-804. doi: 10.16183/j.cnki.jsjtu.2018.232
	JIANG Xiaokang, ZHANG Peng, LÜ Youlong, et al. Hybrid ant colony algorithm for batch scheduling in semiconductor furnace operation[J]. Journal of Shanghai Jiao Tong University, 2020, 54(8): 792-804.
[16]	曾博, 穆宏伟, 董厚琦, 等. 考虑5G基站低碳赋能的主动配电网优化运行[J]. 上海交通大学学报, 2022, 56(3): 279-292. doi: 10.16183/j.cnki.jsjtu.2021.367
	ZENG Bo, MU Hongwei, DONG Houqi, et al. Optimization of active distribution network operation considering decarbonization endowment from 5G base stations[J]. Journal of Shanghai Jiao Tong University, 2022, 56(3): 279-292.
[17]	闫龙川, 白东霞, 刘万涛, 等. 人工智能技术在云计算数据中心能量管理中的应用与展望[J]. 中国电机工程学报, 2019, 39(1): 31-42.
	YAN Longchuan, BAI Dongxia, LIU Wantao, et al. Application and prospect of artificial intelligence technology in energy management and optimization for cloud computing data center[J]. Proceedings of the CSEE, 2019, 39(1): 31-42.
[18]	杨楠, 叶迪, 林杰, 等. 基于数据驱动具有自我学习能力的机组组合智能决策方法研究[J]. 中国电机工程学报, 2019, 39(10): 2934-2946.
	YANG Nan, YE Di, LIN Jie, et al. Research on data-driven intelligent security-constrained unit commitment dispatching method with self-learning ability[J]. Proceedings of the CSEE, 2019, 39(10): 2934-2946.
[19]	NAIR V, BARTUNOV S, GIMENO F, et al. Solving mixed integer programs using neural networks[DB/OL]. (2021-07-29)[2023-12-05]. https://arxiv.org/abs/2012.13349.
[20]	DEY S, DUBEY Y, MOLINARO M, et al. A theoretical and computational analysis of full strong-branching[DB/OL]. (2021-11-09)[2023-12-05]. https://arxiv.org/abs/2110.10754.
[21]	陈敏, 高赐威, 陈宋宋, 等. 考虑数据中心用电负荷调节潜力的双层经济调度模型[J]. 中国电机工程学报, 2019, 39(5): 1301-1314.
	CHEN Min, GAO Ciwei, CHEN Songsong, et al. Bi-level economic dispatch modeling considering the load regulation potential of internet data centers[J]. Proceedings of the CSEE, 2019, 39(5): 1301-1314.
[22]	CHEN Z, WU L, LI Z. Electric demand response management for distributed large-scale internet data centers[J]. IEEE Transactions on Smart Grid, 2017, 5(2): 651-661. doi: 10.1109/TSG.2013.2267397 URL
[23]	KNUEVEN B, OSTROWSKI J, WATSON J P. On mixed-integer programming formulations for the unit commitment problem[J]. INFORMS Journal on Computing, 2020, 32(4): 857-876.
[24]	ZHUO Z, DU E, ZHANG N, et al. Cost increase in the electricity supply to achieve carbon neutrality in China[J]. Nature Communications, 2022, 13(1): 1-13. doi: 10.1038/s41467-021-27699-2

算力枢纽节点	功能规划
1—京津冀枢纽	辐射华北、东北乃至全国,引导业务向西部迁移
2—粤港澳枢纽	辐射华南乃至全国,引导业务向西部迁移
3—长三角枢纽	协同长三角算力辐射全域,引导业务向西部迁移
4—成渝枢纽	平衡城市间算力资源部署,衔接“东数西算”
5—贵州枢纽	主要支持长三角与粤港澳大湾区算力需求
6—甘肃枢纽	重点服务京津冀、长三角、粤港澳大湾区
7—内蒙古枢纽	为京津冀、长三角提供算力保障
8—宁夏枢纽	发挥可再生能源优势,承接东部算力需求

参数	取值
学习率,Learning rate	0.001
训练周期尺寸,Epoch size	312
训练批尺寸,Batch size	40
终止对偶间隙/%	2
求解时间上限/s	1 200

对比方法	平均求解时间/s	所提方法求解时间平均缩短比例/%
RPB	2 931	39.1
CPLEX	4 450	66.7
Extratrees	2 821	13.5
FSB	—	—
CPLEX(并行)	4 450	66.7
CPLEX(串行)	2 328	38.1

调度周期/min	碳排放成本/元	冷却成本/元	服务成本/元	总成本/元
60	17 447	105 842	318 108	441 397
30	17 227	105 316	317 045	439 588
15	17 125	105 417	316 675	439 217