| In recent years,with the rapid development of bionics,micromechanics and electronics,modern control and other disciplines,the bionic flapping-wing aerial vehicle(FWAV),as the bionic robot that mimics biological flight,has received widespread attention from researchers at home and abroad.Compared with the traditional fixed-wing unmanned aerial vehicle(UAV)and quadrotor UAV,the FWAV has several characteristics such as: low noise,stealth and maneuverability,which will be widely used in the military field and civil field in the future.However,it is susceptible to many problems such as external disturbance,actuator failure and model uncertainty during flight,forcing its tracking stability to a predetermined trajectory to be reduced and even crashes.Therefore,in this paper,corresponding anti-disturbance controllers and fault-tolerant controllers are designed for FWAV models with external disturbance,actuator failures and uncertainties,combining adaptive control and neural network control.The specific research contents are as follows:First of all,the flight mechanism of FWAV is studied,and the coordinate system suitable for flapping-wing aircraft is established,The mathematical models of attitude subsystem and position subsystem of FWAV are established in Lagrangian form,which lays the foundation for the later control method.Secondly,Based on the established FWAV system model,considering the external disturbance and the uncertainties of internal parameters of the model,neural network is designed to approximate and to estimate the complex nonlinearity and uncertainties in the mathematical model of FWAV.Meanwhile,this paper uses adaptive technology to counteract the adverse effects of external disturbance.Based on the backstepping recursive framework,it can design a dynamic surface controller by introducing a first-order filter,which overcomes the limitation of "differential explosion" in traditional backstepping recursive design.Then,a Lyapunov function is proposed to prove the closed-loop system stability and the semi-global uniform ultimate boundedness of all state variables.The results show that the proposed controller can not only make the steady-state error converge better,but also improve the convergence speed.Simulation results indicate that the control method can effectively deal with uncertainties and external disturbances and can track the desired trajectories well.Finally,the fault-tolerant control problem of FWAV with actuator failure is further studied under the condition of considering the above external disturbances and the uncertainty of internal parameters of the model.Actuator failure will affect the tracking performance of FWAV position and attitude.In order to improve the fault-tolerant ability of FWAV,a fault-tolerant control scheme based on nonsingular fast terminal sliding mode is proposed.In order to eliminate the influence of actuator failure and uncertainty on system performance in the model,radial basis function neural network is used to compensate actuator failure and uncertain dynamics,and a nonsingular fast terminal sliding mode is constructed to further improve the stability of the system and accelerate the recovery of system stability after failure.In addition,considering that it is difficult to measure the angular velocity of FWAV,a state observer is designed to estimate the angular velocity of FWAV,and neural network is combined to improve the estimation accuracy.Finally,it is proved by Lyapunov method that the designed controller can ensure the asymptotic stability of the attitude and position closed-loop system,and the effectiveness of the control scheme is verified by Matlab numerical simulation. |
