This work focuses on the attitude control problem for hypersonic morphing vehicles(HMVs) with uncertainties. A span-morphing HMV model is firstly established with lumped disturbances. Based on fixed-time technique, a disturbance observer with fuzzy logic system(FLS) is proposed to enhance the robustness by estimating the unmodelled dynamics and external disturbances in fixed settling time. The multivariable fixed-time sliding mode manifold is applied to the transformed control system, which guarantees a direct control input design from attitude command and decent tracking performance. Then, the control system degenerates into an autonomous system, and the closed-loop fixed-time stability is ensured via Lyapunov synthesis and homogeneity theory. Finally, simulation results are presented to demonstrate the effectiveness of the proposed control scheme.

This paper develops a fuzzy disturbance observer-based fixed-time control technique to solve the attitude-tracking problems for hypersonic morphing vehicles (HMVs). According to the dynamic model of HMVs, the control-oriented model is established to simplify the design procedure. Fuzzy fixed-time disturbance observers are proposed for high-precision and fast disturbance estimation. The attitude controller is developed based on multivariable sliding mode control and fixed-time stability technique. Compared to the conventional fixed-time control methods, this paper adopts variable exponent coefficients to achieve fixed-time stability. The controller will be obtained without lengthy inner and outer loop design procedures. The closed-loop stability is proved via Lyapunov synthesis. Finally, several scenarios are designed to assess performance.

An onboard three-dimensional (3D) trajectory generation approach based on the reinforcement learning (RL) algorithm and deep neural network (DNN) is proposed for hypersonic vehicles in glide phase. Multiple trajectory samples are generated offline through the convex optimisation method. The deep learning (DL) is employed to pre-train the DNN for initialising the actor network and accelerating the RL process. Based on the offline deep policy deterministic actor-critic algorithm, a flight target-oriented reward function with path constraints is designed. The actor network is optimised by the end-to-end RL and policy gradients of the critic network until the reward function converges to the maximum. The actor network is considered as the onboard trajectory generator to compute optimal control values online based on the real-time motion states. The simulation results show that the single-step online planning time meets the real-time requirements of onboard trajectory generation. The significant improvement in terminal accuracy of the online trajectory and the better generalisation under biased initial states for hypersonic vehicles in glide phase is observed.

The attitude-tracking problem of hypersonic morphing vehicles (HMVs) is investigated in this research. After introducing variable-span wings, the optimal aerodynamic shape is available throughout the entire flight mission. However, the morphing wings cause significant changes in aerodynamic coefficients and mass distribution, challenging the attitude control. Therefore, a complete design procedure for the flight control system is proposed to address the issue. Firstly, the original model and the control-oriented model of HMVs are built. Secondly, in order to eliminate the influence caused by the multisource uncertainties, an adaptive fixed-time disturbance observer combined with fuzzy control theory is established. Thirdly, the fixed-time control method is developed to stabilise hypersonic morphing vehicles based on a multivariable sliding mode manifold. The control input can be obtained directly. Finally, the effectiveness of the proposed method is proved with the help of the Lyapunov theory and simulation results.

An event-triggered practical prescribed time control (PPTC) strategy is proposed for hypersonic morphing vehicles (HMVs) with system uncertainties and limited computation resources. HMVs can frequently change the aerodynamic profile to keep the desired aerodynamic properties. Attitude stabilization is essential for HMVs in continuous morphing stages. First of all, the HMV model is established and investigated to support the control system design. As a result, the complete morphing process will lead to significant improvements in aerodynamic coefficients, requiring fast-tracking and adaptation. Then, an appointed-time prescribed performance function (ATPPF) is proposed. Based on the ATPPF and the HMV model, the PPTC technique is induced for HMVs. Thirdly, the event-triggered PPTC is presented based on morphing information-driven (MI-driven) events, which are a part of the actual flight status. The developed MI-driven event-triggering mechanism could lead to a lower triggering number than the conventional time-triggering mechanism (TTM) and the commonly used relative-threshold event-triggering mechanism (RTETM). Finally, the prescribed time stability is theoretically proven via Lyapunov thesis while excluding the Zeno behavior. The simulations validated the effectiveness and the properties of the proposed control scheme for HMVs.

Large-scale morphing poses a great challenge to the safety and stability of the control system for high-speed morphing vehicles (HMV). Although reinforcement learning (RL) has significant potential to address the high uncertainty and strong nonlinearity of HMV control, it cannot strictly guarantee the safety of the system. In this paper, a robust and safe RL method is proposed for HMV attitude control in a hybrid control framework that synergizes a data-driven RL controller with a model-driven safety filter. The method starts with a model-free RL called proximal policy optimization (PPO) combined with gated recurrent units (GRU), which is then trained to learn a controller in an environment with domain randomization. Next, based on system state constraints, a robust higher-order control barrier function (HOCBF) is introduced and integrated with an adaptive disturbance observer (DOB) to create a safety filter. To simplify the solving process, the safety filter hierarchically eliminates unsafe outputs generated by the RL-based controller according to the priority of state constraints. Finally, comparative simulations and Monte Carlo simulations are performed to verify the effectiveness, robustness and generalization of the proposed method.