无人机集群对抗决策算法研究综述
Review of UAV Swarm Air-combat Decision-making Algorithms
李潍 1黄诗怡 1刘宏明 1孙张俊2
作者信息
- 1. 东南大学,江苏 南京 210096
- 2. 航空工业西安飞行自动控制研究所,陕西 西安 710076
- 折叠
摘要
无人机集群博弈对抗已经成为未来战争的发展趋势,无人机对抗决策算法的选择对提升无人机集群作战能力至关重要.本文深入探讨了基于规则的、基于博弈论的和基于神经网络的三大类无人机集群博弈对抗决策算法,并对它们的优势和局限性进行了全面分析与总结.在此基础上,提出将"基于多智能体强化学习的信用分配模型"和"基于角色的多智能体强化学习模型"应用于无人机集群博弈对抗的研究思路.最后,强调了选择适当的决策算法对于提高无人机集群作战效能的重要性,并为未来无人机对抗决策的发展提出了有益的建议,为相关领域的研究和应用提供了深入见解.
Abstract
UAV swarm air-combat has become the development trend of future warfare,and the selection of UAV swarm air-combat decision-making algorithms is crucial for improving the UAV swarm combat ability.This paper delve into three types of UAV swarm air-combat decision-making algorithms based on rules,game theory,and neural networks,and comprehensively analyze and summarize their advantages and limitations.On this basis,this paper propose to apply the multi-agent reinforcement learning based credit assignment model and role-based malti-agent reinforcement learning model and design for UAV swarm air-combat.Finally,it emphasize the importance of selecting appropriate decision algorithms to improve the combat effectiveness of UAV clusters,and provide useful suggestions for the development of UAV countermeasures decision-making in the future,providing in-depth insights for research and application in related fields.
关键词
无人机集群/博弈对抗/专家系统/博弈论/多智能体强化学习Key words
UAV swarm/air-combat/expert system/game theory/multi-agent reinforcement learning引用本文复制引用
基金项目
航空科学基金(20200058069001)
出版年
2024
国家科技期刊平台
NSTL
万方数据