基于多智能体强化学习的勤务保障指挥调度方法

Aircraft Commanding and Scheduling Method of Flight Deck Based on Multi-Agent Reinforcement Learning

在时、空、资源受限的复杂勤务保障场景中,实现动态、高效且可靠的勤务保障组织指挥调度,对于提升大型舰船平台飞机出动能力至关重要。通过分析飞机勤务保障指挥调度任务特点,建立多机、多保障作业并行执行的勤务保障指挥调度马尔科夫决策过程,将多智能体技术与强化学习深度结合,构建多智能体深度确定性策略迭代模型,自动生成保障计划。经仿真实验验证,所提出的勤务保障指挥调度方法,能有效满足飞机勤务指挥调度优化需求。

In highly complex aircraft supporting process on flight deck,it is critical to achieve a fast-response and reliable scheduling to improve shipboard launch capability.In this paper,we analyze the characteristics of the aircraft supporting missions on flight deck,and construct a Markov decision process(MDP)for the aircraft commanding and scheduling.By deeply combining multi-agent technologies with reinforcement learning,it enables efficient multi-aircraft parallel scheduling scheme optimization and generation.Simulation results show that the proposed method satisfies the requirement of dynamic aircraft supporting optimization.