航空学报 > 2026, Vol. 47 Issue (2): 332170-332170   doi: 10.7527/S1000-6893.2025.32170

数据-知识双驱动的编队目标意图识别方法

刘祥雨1, 王刚1, 王思远1(), 陈卓文2   

  1. 1.空军工程大学 防空反导学院,西安 710051
    2.空军工程大学 航空工程学院,西安 710051
  • 收稿日期:2025-04-27 修回日期:2025-05-15 接受日期:2025-06-12 出版日期:2025-06-30 发布日期:2025-06-27
  • 通讯作者: 王思远 E-mail:982268878@qq.com
  • 基金资助:
    国家自然科学基金(62402521)

A data-knowledge dual-driven method for formation target intention recognition

Xiangyu LIU1, Gang WANG1, Siyuan WANG1(), Zhuowen CHEN2   

  1. 1.Air and Missile Defense College,Air Force Engineering University,Xi’an 710051,China
    2.Aeronautical Engineering College,Air Force Engineering University,Xi’an 710051,China
  • Received:2025-04-27 Revised:2025-05-15 Accepted:2025-06-12 Online:2025-06-30 Published:2025-06-27
  • Contact: Siyuan WANG E-mail:982268878@qq.com
  • Supported by:
    National Natural Science Foundation of China(62402521)

摘要:

复杂对抗下的防空作战中,如何对战场目标的意图进行识别成为防空作战态势认知的关键,针对编队目标意图识别过程中的状态变化灵活、意图规律性强等特点,提出了数据-知识双驱动的编队目标意图识别方法,首先设计了数据-知识双驱动的意图识别总体框架,采取串联结合方式解决单一方法的局限性。其次基于该框架设计数据驱动的基于双向长短时记忆网络和条件生成对抗网络的目标状态预测(BC-TSP)方法,解决知识驱动全局搜索能力不足的问题。而后提出了知识驱动的基于模糊决策树的意图推理方法,解决数据驱动解空间爆炸、可解释性低的问题。最后分别进行实验验证,结果分析表明本方法既有着优秀的时序特征数据学习能力,能准确预测编队目标未来的状态信息,又可以根据实际情况高效利用不完备的先验知识,通过可解释性强的推理方式获得编队目标的最终意图,满足防空作战指挥员的态势认知需求,辅助其更好地进行决策。

关键词: 编队目标, 意图识别, 知识驱动, 数据驱动, 模糊决策树

Abstract:

In complex adversarial air defense operations, the identification of battlefield target intent has become crucial for situational awareness. To address the characteristics of flexible state variations and strong regularity in the intent of formation targets, this paper proposes a data-knowledge dual-driven method for formation target intent recognition. First, a hybrid framework of data-knowledge dual-driven intent recognition is designed to address the limitations of using either approach alone. Subsequently, based on this framework, a data-driven Target State Prediction (BC-TSP) method is proposed to solve the problem of insufficient knowledge-driven global search capability, combining Bidirectional Long Short-Term Memory (BiLSTM) and Conditional Generative Adversarial Network (CGAN). Furthermore, a knowledge-driven intent inference method based on fuzzy decision trees is introduced to solve the problem of data-driven solution space explosion and low interpretability. Comprehensive experimental validation demonstrates that the proposed method not only exhibits strong temporal feature learning capability, enabling accurate prediction of future formation target states, but also efficiently leverages incomplete prior knowledge through a highly interpretable reasoning process to determine the final intent of formation targets. This approach satisfies the situational awareness needs of air defense commanders, assisting them in making better-informed tactical decisions.

Key words: formation target, intent recognition, knowledge-driven, data-driven, fuzzy decision tree framework

中图分类号: