| 引用本文: | 陈浩,黄健,刘权,周思航,张中杰.自主空战机动决策技术研究进展与展望[J].控制理论与应用,2023,40(12):2104~2129.[点击复制] |
| CHEN Hao,HUNAG Jian,LIU Quan,ZHOU Si-hang,ZHANG Zhong-jie.Review and prospects of autonomous air combat maneuver decisions[J].Control Theory and Technology,2023,40(12):2104~2129.[点击复制] |
|
| 自主空战机动决策技术研究进展与展望 |
| Review and prospects of autonomous air combat maneuver decisions |
| 摘要点击 1578 全文点击 464 投稿时间：2023-04-12 修订日期：2023-11-25 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 10.7641/CTA.2023.30210 |
| 2023,40(12):2104-2129 |
| 中文关键词 自主空战 机动决策 强化学习 博弈论 数据驱动 优化理论 |
| 英文关键词 autonomous air combat maneuver decision reinforcement learning games data driven optimization theory |
| 基金项目 国家自然科学基金项目(62006237)资助. |
|
| 中文摘要 |
| 在强对抗、高动态的现代空战环境中, 以空战决策为核心的自主空战在减轻飞行员决策压力、提高决策效
率和准确性、提升无人机作战效能、实现有人/无人协同、探索新战术战法等方面发挥着重要作用. 首先, 本文分析
了视距内空战和超视距空战在现代空战中的地位和关系, 给出了自主空战机动决策的定义, 并阐述了其在空战中的
实际意义. 其次, 分析了自主空战机动决策在现代空战发展中的重要作用. 然后, 将现有自主空战机动决策建模与
求解方法分为基于博弈论的方法、基于优化理论的方法和基于数据驱动的方法3大类, 并系统地梳理总结了各类方
法的研究分支、特点、优势及缺陷. 进一步地, 以1v1空战为对抗场景并选取代表算法, 直观对比了3类方法的特点和
性能差异. 最后, 针对当前自主空战机动决策技术的痛点问题提出了未来研究和发展的方向. |
| 英文摘要 |
| In the realm of modern air combat, characterized by intense confrontation and high dynamics, autonomous
air combat plays a crucial role. It can alleviate the decision-making pressure on pilots, enhance decision-making efficiency
and accuracy, improve the effectiveness of unmanned aerial vehicles, facilitate manned and unmanned coordination, and
explore new tactics. This paper begins by analyzing the status and interrelationships of within visual range and beyond
visual range air combat. We provide a definition of maneuver decisions and highlighting their significance in modern air
combat. Next, the importance of autonomous air combat maneuver decisions in the advancement of modern air combat
is thoroughly analyzed. Then, existing modeling frameworks and solving methods of autonomous air combat maneuver
decisions are reviewed. These methods are categorized into game theory-based, optimization theory-based, and data-driven methods. Each category is thoroughly summarized, including research branches, characteristics, advantages, and
limitations. Furthermore, taking the 1v1 air combat as the confrontation scenario and selecting representative algorithms,
the performance differences of the three category methods are compared intuitively. Finally, taking into account the core
issues, we present several directions for future research and development. |
|
|
|
|
|