Recurrent-Functional-Link-Network-Based Predictive Control Hypersonic Vehicles with Dynamical Uncertainties

This paper presents a novel adaptive predictive control methodology for air-breathing hypersonic vehicles (AHVs) in the presence of dynamical thrust and parameter uncertainties. The method combines the nonlinear generalized predictive control (NGPC) with a new recurrent-functional-link- network (RFLN) adaptive law proposed here. The RFLN is a simple recurrent neural network without hidden layers, so it is suitable for the approximation of dynamical uncertainties during the hypersonic flight. The RFLN weights are online tuned by the derived adaptive laws based on Lyapunov stability theorem for the first time. The learning process does not need any offline training phase. Additionally, a robust controller whose gain can be adaptively adjusted is designed to recover the residual of the approximation error. Finally, simulation results show better performance of the controller for the AHV attitude tracking than the method compared, moreover, the robustness to dynamical parameters variations and the disturbance rejection are successfully accomplished.