Intelligent Adaptive Control Of Hypersonic Vehicles With Actuator Constraints

The air-breathing hypersonic vehicle has important research significance in the aerospace and military fields because of its high flight speed,strong maneuverability,and high cost-effectiveness.In recent years,many hypersonic vehicle projects have made great progress.However,due to the existence of strong nonlinearities,strong couplings,the uncertainties of aerodynamic parameters,and external disturbances,along with constrained actuators,the control system design still faces huge challenges.This paper focuses on the tracking control problem of the longitudinal channel of the air-breathing hypersonic vehicle,and mainly conducts the following research.Firstly,the control-oriented model of the longitudinal elastic air-breathing hypersonic vehicle is obtained.On the one hand,the elastic longitudinal model is obtained,where the expressions of aerodynamic force,torque and elastic mode are obtained by polynomial curve fitting.On the other hand,servo system models of constrained actuators considering transfer functions,output amplitude and rate limits are established,.Secondly,the tracking control problem of air-breathing hypersonic vehicles with uncertain aerodynamic parameters is studied.For the velocity tracking problem,an adaptive dynamic inverse controller is designed.For the altitude tracking problem,an adaptive backstepping controller is designed.The designed adaptive controller also uses projection functions to avoid singularity problems.Thirdly,the tracking control problem of hypersonic vehicles with constrained actuators is studied.Velocity tracking control is based on adaptive dynamic inversion,and an auxiliary signal is designed to deal with the constrained output of the scramjet.The altitude tracking control is based on the adaptive backstepping method.First,the primary control command of control surfaces is obtained,and then the model predictive control method is used to modify the command to handle constrained actuators.In addition,this paper also uses neural networks to reduce the complexity of adaptive update laws and adaptive methods to estimate equivalent disturbances,and therefore,avoids the chattering problem.Fourthly,the reinforcement learning-enhanced adaptive tracking control of the airbreathing hypersonic vehicle with constrained actuators is studied.Considering the existence of external disturbances,an adaptive controller based on reinforcement learning is designed.The adaptive super-twisting differential operator is used to estimate the uncertainty of aerodynamic parameters and external disturbances,and the estimation error of the operator is handled by adaptive methods,which reduces the complexity of adaptive update laws.From the view of control allocation,the reinforcement learning is used to replace the model predictive control to deal with constrained actuators,and auxiliary signals are designed to handle the possible control allocation error,which avoids solving the online optimization problem and improves the calculation efficiency.The designed control systems in this paper have been proved to be stable by the Lyapunov method,and the validity and robustness of these controllers have been verified by simulations.Finally,the research is summarized and the future work is stated at the end of this paper.