As the capability of the anti-missile interception system continues to strengthen, the penetration capability of ballistic missiles is crucial for completing combat missions. This paper proposes a multi-missile penetration strategy based on deep reinforcement learning (DRL), which improves the penetration capability of ballistic missiles from two aspects: the number of ballistic missiles and the maneuvering penetration policy. In the proposed method, the penetration policy is modeled as a deep neural network, and its parameters are trained under a reinforcement learning (RL) framework. In the training process, a centralized training and decentralized execution strategy is adopted. All missiles share a maneuver penetration policy network, and use it to collect interactive experiences, and then the shared maneuver penetration policy network is trained by the experience collected by all missiles. The simulation results show that the trained maneuver penetration policy network can enable the ballistic missiles to successfully penetrate the defense of the interceptors.
Learning-Based Multi-missile Maneuver Penetration Approach
Lect. Notes Electrical Eng.
International Conference on Autonomous Unmanned Systems ; 2022 ; Xi'an, China September 23, 2022 - September 25, 2022
Proceedings of 2022 International Conference on Autonomous Unmanned Systems (ICAUS 2022) ; Kapitel : 348 ; 3772-3780
2023-03-10
9 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
High-speed Offensive Missile Evasive maneuver
AIAA | 1986
|Target maneuver detection and estimation (missile guidance)
Tema Archiv | 1988
|