基于再励学习的歼击机安全着陆横侧向协调控制

基于大系统递阶控制思想,提出了一种运用再励学习算法设计歼击机自动着陆横侧向协调控制系统的方法.将再励学习算法引入歼击机着陆阶段的横侧向子系统的协调控制中,对滚转通道和航向通道的控制器参数进行在线寻优,提高了在线补偿神经网络消除非线性系统逆误差的能力,有效地改善了歼击机着陆阶段横侧向控制子系统的鲁棒性能.通过将这种方法应用于某型歼击机自动着陆控制的实时仿真研究,仿真结果表明了本文方法的有效性和可行性.

Fighter safe landing lateral-directional control based on reinforcement learning

A fighter safe landing lateral-directional control method is presented based on reinforcement learning algorithm(RL),using hierarchical control of dynamical large-scale systems theory.The method,searching optimized controllers' parameters of rolling channel and yawing channel by applying RL algorithm in the lateral-directional coordination control of fighter,is used to regulate the error in the plant inversion due to modeling uncertainties and disturbances,and effectively improves the robustness of fighter lateral-directional control subsystem during the landing phase.The application of the method in safe landing control of fighter is studied by the real-time simulation.Results show that the method is effective and feasible.