摘要
针对调度算法无法动态适应数据中心状态动态变化和用户需求多样化的问题,提出一种基于近端策略优化的数据中心两阶段任务调度算法。通过设计优先级函数为任务提供优先级,采用近端策略优化方法适应数据中心状态动态变化和用户需求的多样化。在任务选择阶段通过计算任务的优先级,优先调度高优先级任务;在物理服务器选择阶段,智能体根据实时的数据中心状态和用户需求,灵活地调整任务调度决策,实现资源的高效分配。实验结果表明,该算法性能优于现有的启发式算法以及常用强化学习算法。
Aiming at the problem that the scheduling algorithm cannot dynamically adapt to the dynamic changes of the data center status and the diversified user needs,a two-stage task scheduling algorithm for the data center based on proximal policy optimization was proposed.The priority for tasks was provided by designing a priority function,and the proximal policy optimization method was adopted to adapt to the dynamic change of data center state and the diversification of user needs.In the task selection stage,the priority of the task was calculated and the high-priority tasks were prioritized.In the physical server selection stage,the agent flexibly adjusted task scheduling decisions based on real-time data center status and user needs to achieve efficient resource allocation.Experimental results show that the performance of the proposed algorithm is better than that of the existing heuristic algorithms and commonly used reinforcement learning algorithms.
作者
徐涛
常怡明
刘才华
XU Tao;CHANG Yi-ming;LIU Cai-hua(Key Laboratory of Civil Aviation Smart Airport Theory and System,Civil Aviation University of China,Tianjin 300300,China;School of Computer Science and Technology,Civil Aviation University of China,Tianjin 300300,China)
出处
《计算机工程与设计》
北大核心
2025年第3期712-718,共7页
Computer Engineering and Design
基金
天津市教委科研计划基金项目(2021KJ037)
面向军民航协同运行的智慧空管信息标准与技术研究基金项目(2021YFF0603902)
民航航空公司人工智能重点实验室自主课题基金项目(CZAILAB-COO-KJAI20001)。
关键词
调度算法
数据中心
任务调度
强化学习
近端策略优化
优先级
两阶段
scheduling algorithm
data center
task scheduling
reinforcement learning
proximal policy optimization
priority
two-stage
