1 State Grid Shanghai Energy Interconnection Research Institute Co., Ltd., Shanghai 201210, China
2 State Grid Hubei Electric Power Research Institute, Wuhan 430072, China
3 Institute of Electrical Engineering, Chinese Academy of Sciences, Beijing 100190, China
Show less
文章历史+
Received
Published
2025-03-12
2026-01-08
Online Date
2026-05-27
PDF (9125K)
摘要
近年来,人工智能、分布式计算等技术极大推动了配电网运行性能的提升,但在任务计算、数据处理效率等方面仍面临巨大挑战。为此,提出一种基于深度强化学习(deep reinforcement learning, DRL)的边缘协同任务卸载博弈优化策略,以提升配电网任务执行效率和系统整体计算性能。首先,建立了考虑任务优先级的配电网边-边协同任务计算模型。进一步,将任务卸载问题建模为Stackelberg博弈。其中,本地服务器作为领导者,以优化任务执行时延和计算成本的综合性能为目标;目标卸载服务器作为跟随者,以提供计算服务获取最大收益为目标。然后,证明了博弈模型纳什均衡的存在性。最后,提出了基于多智能体深度确定性策略梯度(multi-agent deep deterministic policy gradient, MADDPG)的博弈模型求解方法。算例结果表明,与随机卸载和贪心卸载算法相比,所提方法下本地服务器效用值分别提高了66.7%和52.4%。任务的平均执行时间比随机卸载减少14.4%。同时,配电网各服务器的计算负荷分配更加均衡,平均负荷系数较随机卸载方法降低了11.9%。
Abstract
In recent years, technologies such as artificial intelligence and distributed computing have significantly enhanced the operational performance of power distribution grids. However, significant challenges remain in areas such as computational task execution and data processing efficiency. An edge collaborative task offloading game optimization strategy based on deep reinforcement learning (DRL) was proposed to enhance task execution efficiency and overall computational performance for distribution networks. Firstly, a distribution network edge to edge collaborative task calculation model considering task priority was established. Furthermore, the task offloading problem was modeled as a Stackelberg game. Where the local server, as the leader, aims to optimize the comprehensive performance of task execution delay and computing cost. As a follower, the target uninstall server aims to provide computing services to maximize revenue. Then, the existence of Nash equilibrium in game model was proved. Finally, a game model solving method based on multi-agent deep deterministic policy gradient (MADDPG) was proposed. The case study results show that the local server utility value under the proposed method is increased by 66.7% and 52.4% respectively compared with the random offloading and greedy offloading algorithms. The average execution time of tasks is 14.4% less than that of random offloading. Furthermore, the load distribution of servers in the distribution network is more balanced, and the average load coefficient is reduced by 11.9% compared with the random unloading method.
TAN Chuan-yu, WANG Peng, LIN Jia-ying, QIN Si-jun, LEI Yang, YIN Jing-yuan.
Edge Collaborative Task Offloading Game Optimization Strategy for Distribution Networks Based on Deep Reinforcement Learning[J].
Science Technology and Engineering, 2026, 26(1): 181-190 DOI:10.12404/j.issn.1671-1815.2501649
近年来,人工智能(artificial intelligence, AI)、分布式计算及电力物联网(Internet of Things, IoT)等前沿技术迅猛发展,为能源消费的高质量发展提供了强有力支撑[1]。然而,受限于当前配电网计算、通信能力以及数据处理效率等因素,配电网在任务实时响应和处理效率等方面仍面临严峻挑战[2-3]。边缘协同计算通过优化配电网的计算资源分配,能够显著降低任务计算延迟,增强配电网的实时响应能力,提高主动配电网(active distribution network, ADN)的整体运行效率[4]。此外,基于数据驱动的策略无需依赖精确的数学模型,能够灵活应对配电网的复杂变化,并动态调整计算资源分配,以优化系统的整体运行[5]。为此,研究基于数据驱动的配电网边缘协同计算策略是解决配电网计算能力不足问题,提供低时延、高效计算服务的有效技术手段之一。
ChenWeijiang, JinXiaoling, WuMing, et al. Thinking on the rapid evolution of distribution network form under the carbon peaking and carbon neutrality goals[J]. Proceedings of the CSEE, 2024, 44(17): 6811-6817.
LiuKeyan, JiaDongli, LiZhao, et al. Coordinated optimization of active distribution network based on improved analytical target cascading method[J]. Science Technology and Engineering, 2025, 25(2): 621-629.
TongJie, QiZihao, PuTianjiao, et al. Edge intelligence to power internet of things: concept, architecture, technology and application[J]. Proceedings of the CSEE, 2024, 44(14): 5473-5495.
ZhouZhenyu, WangZhao, LiaoHaijun, et al. 5G cloud-edge-end collaboration framework and resource scheduling method in power Internet of Things[J]. Power System Technology, 2022, 46(5): 1641-1651.
LeiAoyu, JiangYoujin, LiuChengxi, et al. Probabilistic power flow calculation based on the data-driven polynomial chaos approximation[J]. Science Technology and Engineering, 2025, 25(2): 598-609.
WangTing, ShaoPeng, LiGuangquan, et al. Improved particle swarm optimization algorithm for cloud computing task scheduling[J]. Science Technology and Engineering, 2023, 23(29): 12594-12603.
[13]
ZhongJ, HuangQ, LiL, et al. Enhancing voltage compliance in distribution network under cloud and edge computing framework[J]. IEEE Transactions on Cloud Computing, 2023, 11(2): 1217-1229.
[14]
LinC, HuB, ShaoC, et al. Computation offloading for cloud-edge collaborative virtual power plant frequency regulation service[J]. IEEE Transactions on Smart Grid, 2024, 15(5): 5232-5244.
[15]
XuS, LiaoB, YangC, et al. Deep reinforcement learning assisted edge-terminal collaborative offloading algorithm of blockchain computing tasks for energy Internet[J]. International Journal of Electrical Power & Energy Systems, 2021, 131: 107022.
[16]
LiuC, WengJ, YangA, et al. Enabling efficient and privacy-preserving aggregation communication and function query for fog computing-based smart grid[J]. IEEE Transactions on Smart Grid, 2020, 11(1): 247-257.
[17]
QinP, WuX, LiuM, et al. Energy-efficient resource allocation for space-air-ground integrated industrial power Internet of Things network[J]. IEEE Transactions on Industrial Informatics, 2024, 20(4): 5274-5284.
[18]
SunY, ZhangY, LiL, et al. Collaborative dynamic task allocation with demand response in cloud-assisted multiedge system for smart grids[J]. IEEE Internet of Things Journal, 2022, 9(4): 3112-3124.
WangZhifu, YangZhongyi, LuoWei, et al. A review of remaining useful prediction for lithium-ion battery based on data-driven[J]. Science Technology and Engineering, 2023, 23(15): 6279-6289.
WuRunze, HuoJinxin, GuoHaobo. DQN-based task offloading strategy for power co-computing and caching[J]. Electric Power Construction, 2024, 45(8): 149-158.
YaoNan, LiuZiquan, QinJianhua, et al. Offloading optimization of the edge computing task based on the power Internet of Things[J]. Science Technology and Engineering, 2022, 22(16): 6577-6584.
[25]
GooiH, WangT, TangY. Edge intelligence for smart grid: a survey on application potentials[J]. CSEE Journal of Power and Energy Systems, 2023, 9(5): 1623-1640.
ChenYuanju, CaiZexiang, CenBoxiong, et al. Deployment and task allocation method of edge computing terminals in distribution network considering task coupling[J]. Automation of Electric Power Systems, 2022, 46(22): 32-40.
[28]
LiuC, ChenL, GaoW. DRL-based online task offloading and energy resource aggregation for edge-computing-empowered smart grid networks[J]. IEEE Internet of Things Journal, 2024, 11(24): 41008-41020.
ZhaoChanchan, GuoXiaomin, HaiXiaowei, et al. Joint optimization strategy of task offloading and resource allocation for cache-assisted mobile edge computing[J]. Science Technology and Engineering, 2023, 23(9): 3812-3819.
KangYiqun, CaiZexiang, ZengXing, et al. An offloading cost based cloud-edge collaborative scheduling method for power distribution network energy management applications[J]. Southern Power System Technology, 2021, 15(9): 61-68.
[33]
ShaoS, SuL, ZhangQ, et al. Multi task dynamic edge-end computing collaboration for urban Internet of Vehicles[J]. Computer Networks, 2023, 227: 109690.