Reinforcement-Learning-Based Finite-Time Fault-Tolerant Formation Control of Fixed-Wing UAVs with Unknown Control Directions
Yi Luo, Ke Zhang, Bo Meng, Bin Jiang · 2023
Firstly, based on reinforcement learning control method, a distributed finite-time formation tactic aims to uncertainties, actuator faults, unsuspected disturbances and unknown control directions is proposed for fixed-wing UAVs. Among them, the actor-critic networks of reinforcement learning strategy are used to approximate uncertainties. Secondly, the command filter technology is introduced into the backstepping design process of the altitude controller. Furthermore, to resolve actuator faults with unknown control directions, this paper utilizes Nussbaum function to handle partial loss of effectiveness and unknown control directions. Besides, this paper employs Lyapunov stability theory to assess the system’s stability and demonstrates that tracking errors converge within a finite timeframe. In the end, simulation outcomes are presented to substantiate the efficacy of the proposed control approach.