基于DQN的多类型拦截装备复合式反无人机任务分配方法
作者:
作者单位:

国防科技大学

作者简介:

通讯作者:

中图分类号:

N945.25

基金项目:

装备发展部领域基金


Task assignment method of compound anti-drone based on DQN for multi type interception equipment
Author:
Affiliation:

National University of Defense Technology

Fund Project:

Field fund of equipment development department

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    针对当前反无人系统单一类型拦截装备无法有效压制无人机的问题,使用多种类型拦截装备,采取最小距离射击原则,构建一种新的复合式反无人机方法,突破多类型拦截装备任务分配问题。针对传统多目标优化算法求解速度慢、智能算法参数难以调整且无法有效平衡全局搜索与局部优化的问题,本文提出一种基于深度Q网络(DQN)的多类型拦截装备复合式反无人机任务分配模型。为了提高算法收敛速度和学习效率,本文方法未采用下一时刻的状态来预测Q值,而是采用当前时刻的状态来预测Q值,同时消除训练过程中Q值过估计的影响。在模型训练过程中针对每个拦截设备采用一对一拦截的方式分别训练对应的智能体,在实际使用时根据最小距离射击原则来决定由满足拦截条件的智能体自主拦截。以国内某机场跑道周围区域开阔地为防护对象,构建反无人机系统的任务分配仿真环境,仿真结果验证了本文方法的有效性。同时,与DQN与Double DQN方法相比,本文改进DQN算法训练的智能体表现更为精确,并且算法的收敛性和所求解的表现更为优异。本文方法为反无人机问题提供新的思路。

    Abstract:

    Aiming at the problem that the single type of intercepting equipment of the current anti-drone system can not effectively suppress the drone, a new compound anti-drone method is constructed by using multiple types of intercepting equipment and adopting the principle of minimum distance shooting to break through the task assignment problem of multiple types of intercepting equipment. Aiming at the problem of slow solution speed of traditional multi-objective optimization algorithm, difficult to adjust the parameters of intelligent algorithm and unable to effectively balance the global search and local optimization, this paper proposes a task assignment model of multi-type interception equipment compound anti-drone based on deep Q network (DQN). In order to improve the convergence speed and learning efficiency of the algorithm, this method does not use the state of the next time to predict the Q value, but uses the state of the current time to predict the Q value, while eliminating the influence of over estimation of Q value in the training process. In the model training process, the corresponding agents are trained for each interception device in a one-to-one interception mode. In actual use, the agents that meet the interception conditions are autonomously intercepted according to the minimum distance shooting principle. The simulation environment of task assignment of anti-drone system is constructed by taking the open area around a domestic airport runway as the protection object. The simulation results verify the effectiveness of this method. At the same time, compared with the DQN and Double DQN methods, the improved DQN algorithm training agent performance is more accurate, and the convergence of the algorithm and the performance of the solution are more excellent. The method in this paper provides new ideas for the anti-drones problem.

    参考文献
    相似文献
    引证文献
引用本文
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2020-06-16
  • 最后修改日期:2020-12-08
  • 录用日期:2021-01-08
  • 在线发布日期:
  • 出版日期: