摘要
为了给时间触发以太网中的事件触发类消息合理地提供时隙、获得更均衡的消息调度时刻,该文提出了基于Q学习的调度规划算法,将消息调度在时间轴上的求解转化为在三维空间上的多宝箱探索问题,实现基于强化学习的网络调度规划算法求解调度时刻表。针对提出的算法进行了仿真实验,并对实验结果进行分析验证,与传统的基于可满足性模理论(SMT)的调度规划算法相比,基于Q学习的调度算法对TTE网络负载均衡性的优化显著超越SMT算法,能更合理地分配网络资源。
In order to provide reasonable time slots for event-triggered messages in time-triggered Ethernet and obtain more balanced message scheduling moments,this paper proposes a Q-learning-based scheduling planning algorithm,which transforms the solution of message scheduling on the time axis into a multi-bucket exploration problem on the three-dimensional space,and implements a reinforcement learning-based network scheduling planning algorithm to solve the scheduling schedule.Simulation experiments are conducted for the proposed algorithm,and the experimental results are analyzed and verified.Compared with the traditional scheduling planning algorithm based on satisfiability mode theory(SMT),the Q-learning-based scheduling algorithm is significantly superior to the SMT algorithm in load balancing for TTE networks and can allocate network resources more rationally.
作者
陈春燕
王红春
王小辉
CHEN Chunyan;WANG Hongchun;WANG Xiaohui(R&D Department,China Academy of Launch Vehicle Technology,Beijing 100076,China;Xi’an Yunwei Zhilian Technology Co.,Ltd.,Xi’an 710025,China)
出处
《实验技术与管理》
CAS
北大核心
2023年第4期52-61,74,共11页
Experimental Technology and Management
关键词
时间触发以太网
强化学习
调度规划
time-triggered Ethernet
reinforcement learning
scheduling planning
