Model-independent Quadrotor Attitude Control Based On Compensation Function Observe

Since the dynamics of the quadrotors UAV is complicated,and is influeneced by various disturbance,it is difficult to build the mathematical model of the quadrotor aircraft accurately,leading to the difficulty of parameter adjustment for the modeldependent controller.To a large extent,the application of model-dependent control algorithms is limited.The model-independent controller needs the observer of model and disturbance.However,due to the problems of its own structure,the existing disturbance observer will have large errors when estimating highly nonlinear unknown model functions and disturbances.Therefore,designing a high-precision modelfunction observer to improve the estimation accuracy and designing a modelindependent control method with ease of parameter adjustment is of great significance for improving control quality of quadrotor attitude system and the development of control algorithm.In this paper,a novel nonlinear compensation function observer is proposed,which greatly improves the estimation accuracy of unknown models and disturbances.Based on the observer,a new model-independent control method is designed and applied to the attitude control system of the quadrotor.The performance of the proposed observer and controller is verified by software-in-the-loop,hardwarein-the-loop and actual flight experiments.The research contents mainly include:On the estimation of unknown model bias and disturbance,structure characteristics of the compensation function observer(CFO)and the extended state observer(ESO)are analyzed in depth in this paper.The reason that the CFO is superior to the ESO is pointed out.In order to solve the problem of linear compensation in CFO,radial basis function(RBF)neural network is used to replace linear filter compensation to further improve the estimation accuracy of nonlinear unknown functions.A compensated function observer with RBF(R-CFO)is proposed.Using Lyapunov function,it is proved the estimation convergence for the R-CFO observer.The simulation experiment analyses show that the estimation accuracy of CFO with R-CFO is higher than that of CFO,and is greatly improved compared with that of ESO.Through the value of estimation of the R-CFO being fed back to the controller,the model compensation control(MCC)being model-independent is proposed.The controller is able to actively compensate the model-bias and disturbance in real time and ensure that the closed-loop system reaches the desired pole and solve the problem of complex parameter adjustment.The closed-loop stability is theoretically proved.Through simulation and experiment,three control methods of R-CFO-based model compensation control,active disturbance rejection control and PID are applied to the attitude system control of the quadrotor,and the comparative analysis is carried out to verify the superiority of the model compensation control strategy.The working principle of the multi-rotor control experimental development platform with the quadrotor autopilot system model is analyzed,which lays a foundation for the efficient practical application of the flight control system.Based on the quadrotor with autopilot,a cascade model compensation controller for attitude loop is designed,and the tracking of the quadrotor to the reference angle is realized through software-in-the-loop,hardware-in-the-loop and experiments.The control performance of the model compensation control strategy in the actual engineering system is verified.