Barrier Lyapunov function based reinforcement learning control for air-breathing hypersonic vehicle with variable geometry inlet

Abstract

Based on barrier Lyapunov functions, a reinforcement learning control method is proposed for air-breathing hypersonic vehicles with variable geometry inlet (AHV-VGI) subject to external disturbances and diversified uncertainties. The longitudinal dynamic for the AHV-VGI is transformed into strict feedback form. Controllers for velocity and altitude subsystems are designed, respectively. Taking advantage of the reinforcement learning strategy, two radial basis function (RBF) neural networks are applied to estimate the “total disturbances” in the flight control system. Actor network is used for generating the estimate of the disturbance. Critic network is used for evaluating the estimation accuracy. Prescribed tracking performances and state constraints can be guaranteed by introducing barrier Lyapunov functions (BLFs). Tracking differentiators are used to generate the derivatives of virtual controllers in the backstepping design process. Simulation results illustrate the effectiveness and advantages of the proposed control strategy.