现代防御技术 ›› 2023, Vol. 51 ›› Issue (6): 87-96.DOI: 10.3969/j.issn.1009-086x.2023.06.011

• 指挥控制与通信 • 上一篇    下一篇

基于路径-博弈混合策略的无人机空战机动决策

张瀚文1, 甘旭升1, 魏潇龙1, 童荣甲2   

  1. 1.空军工程大学 空管领航学院,陕西 西安 710051
    2.中国人民解放军94188部队,陕西 西安 710077
  • 收稿日期:2022-09-16 修回日期:2023-06-21 出版日期:2023-12-28 发布日期:2023-12-29
  • 作者简介:张瀚文(2000-),男,江西上饶人。讲师,学士,研究方向为地面领航。

Research on Air Combat Maneuver Decision-Making of UAVs Based on Path-Game Hybrid Strategy

Hanwen ZHANG1, Xusheng GAN1, Xiaolong WEI1, Rongjia TONG2   

  1. 1.Air Traffic Control and Navigation College, AFEU, Xi'an 710051, China
    2.PLA 94188 Troops, Xi'an 710077, China
  • Received:2022-09-16 Revised:2023-06-21 Online:2023-12-28 Published:2023-12-29

摘要:

针对无人机的自主空战机动决策问题,设计了基于路径-博弈混合策略的决策算法。首先根据无人机飞行控制过程中,水平机动和垂直机动可以解耦的原理,提出了相解耦的自主决策机制,使用路径规划实现水平机动决策,使用博弈理论实现垂直机动决策。为提升决策环境的灵活性,设计了能够自适应调整规划范围和分辨率的动态栅格环境。基于QL算法设计路径规划模型,并使用双Q表学习机制改进算法,有效提升了路径规划质量。基于纳什均衡理论构建垂直机动算法模型,根据不同的态势环境设计了代价计算函数,实现了无人机的垂直机动决策。最后,针对一对一空战对抗情景开展仿真验证,验证了算法的有效性,相对于传统基于三维规划空间下的机动决策,可有效缩短规划耗时,提升规划品质。

关键词: 无人机, 机动决策, Q-learning, 纳什均衡, 空战

Abstract:

In view of the autonomous air combat maneuver decision-making problem of unmanned aerial vehicles (UAVs), a decision-making algorithm based on a path-game hybrid strategy is designed. Firstly, according to the principle that horizontal maneuver and vertical maneuver can be decoupled in the flight control process of UAVs, a decoupled autonomous decision-making mechanism is proposed, which uses path planning to realize horizontal maneuver decision-making and adopts game theory to realize vertical maneuver decision-making. In order to improve the flexibility of the decision-making environment, a dynamic grid environment is designed, which can adaptively adjust the planning range and resolution. The path planning model is designed based on Q-learning (QL) algorithm, and the algorithm is improved by using a double Q-table learning mechanism, which effectively improves the quality of path planning. Based on the Nash equilibrium theory, the vertical maneuver algorithm model is constructed, and the cost calculation function is designed according to different situation environments. Thus, the vertical maneuver decision-making of UAVs is realized. Finally, the simulation of a one-to-one air combat confrontation scenario verifies the effectiveness of the algorithm. Compared with the traditional maneuver decision-making based on three-dimensional planning space, the proposed algorithm can effectively shorten the planning time and improve the planning quality.

Key words: UAV, maneuver decision-making, Q-learning, Nash equilibrium, air combat

中图分类号: