自适应动态规划算法在飞行器追逃中的应用

针对飞行器追逃对抗的二人零和微分对策问题,提出基于数据的积分策略迭代自适应动态规划算法,以求解数学模型未知系统的控制律.该算法利用固定时段内有效的状态和输入信息,建立数据模型,并对其进行基于值函数和控制策略的算法迭代,在平面拦截系统完全未知的情况下得到追逃双方的近似最优策略.仿真结果表明,所得到的双方控制策略能在有限界内无限接近最优解,验证了所提出算法的有效性.

Application of adaptive dynamic programming algorithm in the pursuit-evasion of aircraft

To solve the problem of two-player zero-sum differential games in the pursuit-evasion of aircraft,a novel approach for obtaining the control laws of a system with unknown mathematic model is proposed using data-based integral policy iteration adaptive dynamic programming (ADP).The algorithm uses available datderailmenta of state and input on fixed time interval to build up the data models.By using them,iterations are conducted based on the value function and control strategies to get the proximate optimal strategies of both under the circumstance of a completely unknown planar interception system.Simulation results show that both control strategies are approximate to their optimal solutions infinitely in a limited range and confirm the effectiveness of the proposed method.