舰载机飞行计划的分层式人机协同决策方法

A hierarchical human-machine-collaborative decision-making method for carrier-based aircraft flight planning

  • 摘要: 【目的】舰载机作为航空母舰的核心作战力量,其出动架次率是衡量航母作战效能的关键指标之一。在甲板空间有限、保障资源受限的作战背景下,如何制定高效、合理的舰载机飞行计划,成为提升舰载机出动效率与整体作战能力的关键问题。以“尼米兹”级航母作战环境为研究背景,围绕舰载机飞行计划的决策问题,提出了一种分层式人机协同决策方法,旨在实现舰载机出动与回收顺序以及保障作业执行顺序的高效协同优化。【方法】首先,构建了一个融合多维关键指标的优先级排序机制,以综合评估舰载机状态与保障作业的紧急性与重要性。在此基础上,结合舰载机层与保障作业层的结构特征,设计了一种分层强化学习算法,以联合优化舰载机及其保障作业的执行顺序,实现全局高效决策。为进一步提升决策质量,引入了指挥员反馈机制,并设计了动态切换函数,以实现智能体与人类指挥员之间的自适应协同。最后,基于“尼米兹”级航母仿真环境开展了大量对比实验。【结果】实验结果表明,所提方法在保障作业总完成时间和等待时间上分别较基线算法平均降低20.9%和31.2%,验证了其在舰载机飞行计划决策中的有效性与实用价值。【结论】研究成果可为多波次舰载机飞行计划决策提供参考。

     

    Abstract: Objectives Carrier-based aircraft serve as the core combat force of an aircraft carrier, and their sortie rate is a key indicator for evaluating the carrier's operational effectiveness. Under operational conditions with limited deck space and constrained support resources, developing efficient and feasible flight plans for carrier-based aircraft becomes a critical challenge for improving sortie efficiency and overall combat capability. Taking the operational environment of Nimitz-class aircraft carriers as the research context, a hierarchical human-in-the-loop decision-making method is proposed to achieve efficient and coordinated optimization of sortie and recovery sequences, as well as support operation execution orders for carrier-based aircraft. Methods A priority ranking mechanism integrating multiple key indicators was first developed to comprehensively assess the status of carrier-based aircraft and the urgency and importance of their support operations. Building upon this, a hierarchical reinforcement learning algorithm was designed, taking into account the structural characteristics of both the aircraft layer and the support operation layer, to jointly optimize the execution sequence of aircraft and their support operations, thereby achieving globally efficient decision-making. To further enhance decision quality, a commander feedback mechanism was incorporated, along with a dynamic switching function to enable adaptive collaboration between the intelligent agent and human commander. Extensive comparative experiments were conducted in a simulated environment based on the Nimitz-class aircraft carrier. Results Experimental results demonstrate that the proposed method reduces the total completion time and waiting time of support operations by an average of 20.9% and 31.2%, respectively, compared to baseline algorithms, thereby validating its effectiveness and practical value in carrier-based aircraft flight planning decisions. Conclusions The results of this study provide valuable insights for supporting decision-making in multi-wave flight planning of carrier-based aircraft.

     

/

返回文章
返回