【摘要】 针对集群无人机编队协同运动过程中遭遇风扰与执行器故障的问题，提出了一种基于强化学习的自适应容错协同控制算法.首先，考虑无人机编队过程中风扰与故障的影响，建立面向控制的模型；其次，基于个体无人机跟踪偏差，构建可同时调节跟踪性能与同步性能的分布式同步跟踪偏差；然后，设计强化学习中的Actor-Critic神经网络，引入分数阶微积分算子，提出强化学习分数阶自适应容错协同控制器.最后，通过Lyapunov稳定性分析与仿真验证，检验所设计控制器的可行性和有效性.更多还原

【Abstract】 In this paper, we investigate the fault-tolerant control problem for swarm unmanned aerial vehicles(UAVs) against wind effects and actuator faults.Also, a reinforcement learning-based fault-tolerant cooperative control strategy is developed to achieve the safety of swarm flight.First, a control-oriented model is established by considering wind effects and actuator faults.Second, based on the individual tracking error of each UAV,the synchronization tracking error is constructed for each UAV by involving individual tracking and synchronization performances.Third, by designing actor and critic neural networks, and using fractional-order calculus, a reinforcement learning-based fractional-order adaptive fault-tolerant cooperative controller is developed for each UAV.Finally, it is shown by Lyapunov stability analyses and simulation results that the developed control scheme can effectively attenuate adverse effects caused by the wind and faults, and that synchronization tracking errors are uniformly ultimately bounded.更多还原