Research On Attitude Control And Compound Control Assignment Method For Reentry Phase Of Space Vehicle

The space vehicle has important research value in the field of aerospace and military.It usually involves the reusability of the vehicle,which creates huge economic value.Therefore,it is one of the focuses of the major aerospace countries.This paper focuses on the attitude control and actuator compound control assignment of reusable reentry vehicle.Firstly,a simplified general model of the reentry attitude system is built.Then the physical characteristics of the aerodynamic rudder and the reaction control system(RCS)are analyzed,and their accurate mathematical models are established.The establishment of the model laid the foundation for in-depth research in this paper.With the development of intelligence and reinforcement learning theory in the field of control,this paper revolves around adaptive dynamic programming(ADP)to design the reentry attitude controller for the vehicle.First,based on the precise model and the principle of time-scale separation,an sliding mode controller based on ADP is devised.The controller adopts a double-loop structure,in which the sliding mode controller is as a main and ADP is as an auxiliary structure to compensate for the shortage of the sliding mode controller.Considering the influence of the learning speed of the ADP structure on the control performance,it proposes an improvement measure to accelerate the learning.Aiming at the unknown of the actual precise model of the vehicle,this paper further designs two attitude controllers based on the incremental model and does not need to rely on the precise model of vehicle.One is a dual heuristic dynamic programming(DHP)controller organized by neural networks,which introduces a new target critic network to eliminate the influence of the critic network on the internal parameters in ordinary DHP;In addition,the incremental model replaces the model network in the conventional DHP,which does not require offline training in advance and is beneficial to reduce the overall training time of the system.The second is an incremental ADP iterative controller built on an incremental model,without the need for complex neural network construction,and the control law for each iteration step is derived based on the optimization principle and ADP ideas.Finally,the effectiveness of these attitude controllers is verified through simulations.Furthermore,for the compound control and distribution of RCS and aerodynamic rudder,a compound control strategy based on dynamic pressure change is designed,which distributes the total expected torque to the two actuators.For RCS,firstly,pulse width modulation(PWM)is used to modulate the continuous torque commands into discrete signals that can be recognized by RCS.Then,this paper designs a scheme based on nonlinear programming for nozzle selection and rudder assignment,and the assignment of RCS and rudder is transformed into optimization problem.Finally,the simulation results show that the compound control allocation strategy is effective.