基于m-best算法与rollout策略的平台-任务关系优化设计方法

Method of platform-task relation optimization design based on m-best algorithm and rollout strategy

在平台-任务关系优化设计中,考虑单个任务资源分配过程中平台资源冗余度对后续任务分配的影响,分析优化设计过程的约束条件,构建了以最大任务执行精度与最小资源冗余度为综合目标函数的平台-任务关系设计的问题模型,并使用基于m-best算法与rollout策略的方法对问题模型进行求解。m-best算法生成单个任务的m个平台分配方案,rollout策略用于均衡不同优先级任务之间的任务执行精度。最后,分别通过特殊算例和一般算例验证所提优化设计方法的优越性,算例的结果表明,使用该优化设计方法能够使高优先权任务的资源冗余度降低,从而使得整体任务执行精度提高。

During the optimization design of platform-task relation,the impact of platform resource redundancy on subsequent tasks allocation during a single task resource allocation is considered.The constraints of the optimization design process are analyzed.A task-platform relation optimization model with maximum task execution accuracy and minimum resource redundancy as synthetic objective function is constructed.The m-best algorithm and rollout strategy are used to solve this problem.The m-best algorithm generates m different assignments for each task and the rollout strategy is used to equalize each task execution accuracy among different priority tasks.Finally,the superiority of the method is verified by a special case and a general case.The results of the simulation case show that the optimization of the proposed method can reduce the redundancy of the high priority task and the overall task execution accuracy is increased.