现代防御技术 ›› 2022, Vol. 50 ›› Issue (6): 68-82.DOI: 10.3969/j.issn.1009-086x.2022.06.009

• 指挥控制与通信 • 上一篇    下一篇

基于进化式决策树的超视距空战机动决策模型

徐安1(), 郑万泽2, 奚之飞1, 党璐菲3   

  1. 1.空军工程大学,航空工程学院,陕西 西安 710038,陕西 西安 710072
    2.空军工程大学,科研学术处,陕西 西安 710072
    3.中国航天科工集团有限公司 第二研究院,北京 100854
  • 收稿日期:2022-03-12 修回日期:2022-07-14 出版日期:2022-12-28 发布日期:2023-01-06
  • 作者简介:徐安(1984-),男,甘肃庆阳人。副教授,博士,研究方向为航空装备作战使用及空战智能决策技术。通信地址:710038 陕西省西安市灞桥区灞陵路一号E-mail: xuankgd@163.com

Improved Evolutionary Decision Tree for BVR Air Combat Decision Making

An XU1(), Wan-ze ZHENG2, Zhi-fei XI1, Lu-fei DANG3   

  1. 1.AFEU, Aviation Engineering School,Shaanxi Xi’an 710038,China, Shaanxi Xi’ an 710072,China
    2.AFEU, Department of Scientific research, Shaanxi Xi’ an 710072,China
    3.The Second Research Academy of CASIC,Beijing 100854,China
  • Received:2022-03-12 Revised:2022-07-14 Online:2022-12-28 Published:2023-01-06

摘要:

针对超视距空战决策问题,提出一种融合战术实施距离和导弹攻击区等先验知识,并具备进化能力的决策树模型。首先,分析空战制胜机理,建立超视距空战态势模型和战术实施距离模型,在此基础上引入超视距空战因子,建立态势评估模型和决策变量集;其次,基于战术实施距离建立超视距战术动作库,作为解空间动作集;第三,构建基于专家知识规则的决策树,并针对决策树参数进化需求,提出一种基于惯性权重动态调整和混沌优化的新型粒子群优化方法,以提高决策规则树进化效率;最后,基于仿真实验给出不同条件下空战机动决策仿真算例并进行了分析。仿真结果表明,提出的进化式决策树能有效融合战术实施距离、导弹攻击区等先验知识,并具备进化能力,能够有效解决超视距空战战术决策问题。

关键词: 战术实施距离, 空战决策, 战术动作库, 专家系统, 决策树, 粒子群优化

Abstract:

Aiming at the decision-making problem of beyond visual range (BVR) air combat, a decision rule tree model that integrates prior knowledge of tactical implementation distance with evolutionary iteration capabilities is established. The mechanism of air combat victory is analyzed, the BVR air combat situation model and the tactical implementation distance model are established, the BVR air combat factors are introduced and the posture assessment model and the set of decision variables are established; the BVR air combat tactical action decomposition and the construction of the action library are implemented; the BVR air combat Decision tree are established, a new particle swarm optimization method based on dynamic adjustment of inertia weights and chaotic optimization is proposed to improve the evolution efficiency realization of the decision rule tree; based on the simulation experiment, the decision effect analysis of the evolutionary decision tree under different conditions is given. The simulation results show that the evolutionary decision tree proposed can effectively integrate a priori knowledge of tactical implementation distance, missile strike zone, etc., and has the evolutionary capability to effectively solve the BVR tactical decision problem.

Key words: tactical implementation distance, air combat decision, tactical maneuvering library, expert system, decision tree, particle swarm optimization

中图分类号: