Research on Missile Intelligent Penetration Based on Deep Reinforcement Learning

Abstract

Penetration capability is one of the most critical measures for assessing the effectiveness of missile systems. In the face of an increasingly sophisticated antimissile system, this article explores the use of artificial intelligence to achieve missile intelligent penetration. The applicability of the deep reinforcement learning method to the missile intelligent penetration problem is elaborated. The Actor-Critic deep reinforcement learning algorithm is studied. The deep reinforcement learning environment model for the missile penetration scenario is developed based on the generic combat effectiveness simulation system, WESS. A multi-sample collaborative training method is designed. On this basis, a missile intelligent penetration training system framework based on WESS and deep reinforcement learning is proposed to guide the missile penetration intelligence learning.