LSTM Based Modeling And Control Of An Uncertain Coaxial Dual-rotor UAV

Compared with the traditional single-rotor configuration,the small coaxial dualrotor UAV has the advantages of compact physical structure,high execution efficiency and low cost.It has broad application prospects in military fields such as reconnaissance and strike,and civil fields such as aerial photography and detection.However,it is easily affected by the uncertainty of internal parameters and external disturbances during the execution of the task.Therefore,it is very difficult to accurately establish the dynamic model of the small coaxial dual-rotor UAV.This thesis studies the uncertainty modeling and trajectory tracking control of small coaxial dual-rotor UAV from the aspects of experimental platform design,uncertainty inertial parameter identification and wind disturbance estimation.The main work of the thesis includes:(1)The flight principle and force of the small coaxial dual-rotor UAV are analyzed,and an experimental platform is built.First,a dynamic model of a small coaxial dualrotor UAV is established,and the dynamic characteristics of the UAV are analyzed through open-loop simulation experiments.Secondly,the physical structure design and hardware implementation of the small coaxial dual-rotor UAV experimental platform are completed,and the effectiveness of the experimental platform and data acquisition system is verified by flight experiments.(2)An inertial parameter identification method for small coaxial dual-rotor UAV based on Long-Short Term Memory(LSTM)network is proposed.Firstly,an inertial parameter identification model of a small coaxial dual-rotor UAV is established,and an inertial parameter identification framework based on LSTM is designed.Furthermore,the domain randomization method is used to obtain the characteristics of the training data,and the LSTM network model for inertial parameter identification is obtained by normalizing the control input and flight state data of the UAV and training the model.Finally,simulations verify the effectiveness of the proposed method.(3)The adverse effect of wind field on small coaxial dual-rotor UAV is studied,and an LSTM-based wind field estimation model is developed,a trajectory tracking control under wind field disturbance is designed accordingly.First,the LSTM model for wind disturbance estimation is learned through the training of UAV control input and attitude information data.Simulation experiments verify the effectiveness of the model for estimating three typical wind disturbances.Secondly,the particle swarm optimization algorithm is introduced to optimize the hyperparameters of the model,and the inertia weight update strategy is designed,which effectively improves the efficiency of the algorithm.Finally,an LSTM-based PID control is proposed for the trajectory tracking of the small coaxial dual-rotor UAV against wind disturbance,so that the UAV can track the given target trajectory.The simulation results show that the proposed scheme has strong anti-wind disturbance ability.