关键词: 超高速飞行器, 主动防御, 时变幂次固定时间制导律, 深度确定性策略梯度

Abstract: An intelligent collaborative game guidance method based on deep deterministic strategy gradient is designed to deal with the three-body game problem of high-speed vehicle actively using defender to hit incoming attacker and protect its safety. Firstly, a variable-exponent fixed-time guidance law is designed for the defender, which can autonomously adjust the exponent in real-time based on the convergence status of the line-of sight angle rate to accelerate convergence time, and ensure that the line-of-sight angle rate converges to zero with fixed time. Then, a lure maneuvering guidance law based on deep determinis-tic policy gradient for high-speed vehicle is designed. By applying this lure maneuvering guidance law, the high-speed vehi-cle can lure attacker to maneuver to areas that are convenient for defender to hit attacker, which can minimize miss distance of defender, and significantly reduce required maneuverability of high-speed vehicle and defender. Based on the fixed time guidance law and lure maneuvering guidance law mentioned, an intelligent collaborative game guidance method is obtained. Through the coordinated movement of high-speed vehicle and defender, the weakly maneuvering defender can successfully strike the strongly maneuvering attacker with a high probability. Finally, simulations are carried out to verify the effectiveness and strong generalization ability of the designed intelligent collaborative game guidance method based on deep deterministic policy gradients. The simulation results demonstrate that under different game situations, the winning probability exceeds 90% for the high-speed vehicle and defender.

Key words: high-speed vehicle, active defend, variable-exponent fixed-time guidance law, deep deterministic policy gradient