| Underactuated systems are a class of nonlinear systems that the control input dimension is less than the system’s generalized motion freedom.The motion model of a quadrotor aircraft is a typical underactuated nonlinear system with the features of multiple inputs,multiple outputs and strong coupling.Due to various negative influences from internal and external environments such as random noises,time-varying disturbances,model uncertainties,and actuator failures,it is difficult to use the conventional single control method to design the controller of underactuated quadrotor,while taking account of strong robustness,adaptability,and high motion control accuracy.To this end,this thesis focuses on underactuated quadrotor aircraft,and mainly studies the following aspects:First,when the conventional LQG control method is used to design the trajectory tracking controller of a quadrotor,a error system is needed to establish to transform the original output tracking control problem into a state adjustment one,or a reference system is needed to assist the controller design,which is not convenient for engineering applications.Aiming at this problem,this paper proposes a Gaussian information fusion control method,in which the performance index function,the system state equation and output equation,are all converted into the information equations on the co-state and control input,and then,based on these information equations,the optimal estimation of control input and its information weight are directly derived using the linear information fusion estimation method.The design process is intuitive and simple,which greatly reduces the design complexity of LQG output tracking control.Simulation experiments show that this method can enable the underactuated quadrotor to achieve robust tracking control of position and attitude under the Gaussian random noise disturbance acting on six degrees of freedom.Then,the model predictive control of underactuated quadrotor under airflow disturbance and gust disturbance is studied,and a nonlinear information fusion model predictive control method is proposed,in which the performance index function,the system state equation and the output equation are all converted into the information equations on co-state and control input,and then,based on these information equations,a nonlinear information fusion model prediction controller is derived using the nonlinear information fusion estimation.Different from traditional nonlinear model predictive control,nonlinear information fusion model predictive control has explicit expression of control law and does not need to solve a large number of nonlinear dynamic programming problems.Simulation experiments verify the effectiveness and robustness of this control method.Finally,a further study on the unknown input saturation control of an underactuated quadrotor under unmodeled dynamics,uncertain model parameters,and external disturbances,is performed.An anti-saturation adaptive neural network finite-time backstepping control is proposed.This method not only can handle completely unknown input saturation nonlinearity,but also has strong robustness and adaptability to external disturbances and lumped uncertainties.Then,a strict Lyapunov finite-time stability analysis is given to ensure the finite-time convergence of all closed-loop system signals.Simulation experiments verify the effectiveness of the proposed control method.The research results of this thesis have certain academic value and broad engineering application prospects,and some of the results can also be applied to other systems with underactuated characteristics such as underactuated surface ships,underwater submersibles,mobile robots,underacted cranes,etc,and has better universality. |
