Quantized feedback neuroadaptive control for a non-affine hypersonic flight vehicles system with actuator saturation

This paper investigates the attitude tracking control problem of non-affine nonlinear hypersonic vehicles with communication constraints and actuator saturation, taking into account significant unmodeled dynamics and external disturbances. The novelty of this work lies in the proposed adaptive quantized state feedback control strategy, which leverages radial basis function neural networks (RBFNNs) to mitigate communication burdens between the controller and the actuator. Specifically, the design of the quantizer aims to alleviate this communication burden. The controller begins by addressing the non-affine components of the system using the mean value theorem, transforming the attitude tracking problem of a non-affine hypersonic vehicle, characterized by state quantization, input saturation, unmodeled dynamics, and external disturbances, into an affine nonlinear problem with unknown nonlinear functions, unknown control gains, and bounded disturbances. To address this, an adaptive backstepping method is employed, along with the universal approximation capabilities of RBFNNs to approximate the unknown nonlinearities. Adaptive neural compensation terms, derived from the quantized states, are introduced to ensure bounded quantization errors. Further innovations are presented in the design of an auxiliary system to manage actuator saturation, maintaining the controller's saturation characteristics, and in the use of a second-order command filter to mitigate the "complexity explosion" issue inherent in backstepping methods. Finally, adaptive gains compensate for bounded disturbances, neural network estimation errors, and quantization errors. In the stability analysis, a recursive method is used to prove the boundedness of quantization errors, and Lyapunov stability theory is applied to demonstrate the stability of the proposed quantization feedback adaptive tracking control system. Simulation results indicate that, under various disturbances and aerodynamic parameter variations, the tracking error remains within a range of 0.05°, and the convergence time is kept under 0.5 s, validating the effectiveness and robustness of the proposed approach.