Data-driven Learning Control Based Four-dimensional Trajectory Tracking Control For Aircraft

This dissertation focuses on accurate four-dimensional(4D)trajectory tracking control and adaptive 4D trajectory and attitude control for civil aircraft.The scientific significance of this work is that complex modeling and robustness problems within the modelbased control method are avoided.Instead,only the input and output data and measurable state data from the aircraft's movement control system are needed to design the control law.The research in this dissertation expands the application areas of model-free adaptive control and iterative learning control and provides a new solution for 4D trajectory tracking control.The major elements of this work are summarized as follows:1.A point-to-point iterative learning control method with current-cycle feedback and a point-to-point model predictive control method with variable predictive step length are proposed to solve the initial 4D trajectory tracking problem with the controlled time of arrival between city pairs.The proposed methods make full use of the repeated operation characteristics and historical operation data of intercity flight,integrating feedback control ideas and model predictive control techniques respectively,overcoming the interference of repetitive and non-repetitive wind speeds on the route,and enable the aircraft to reach the designated waypoint at the designated time with high control accuracy.Numerical simulations verify the effectiveness of the proposed method in the initial 4D trajectory tracking control.2.A point-to-point model-free adaptive iterative learning trajectory tracking control method is designed to overcome the difficulty of accurately modeling the aircraft's complex dynamics.This method uses the measurable input and output data in the aircraft motion control system to establish a dynamic linearized data model and combines the initial 4D trajectory tracking control object to derive an unconstrained point-to-point data-driven trajectory tracking control law.In addition,given the thrust limitation of the aircraft,a point-to-point data-driven trajectory tracking control method with constraints is developed.Through numerical simulation,the effectiveness of the proposed method for trajectory tracking,sensitivity to input control,and flexibility to the number of waypoints are analyzed.3.Two data-driven trajectory tracking predictive control methods are designed to meet the requirements of accurate 4D trajectory tracking and trajectory prediction.They are model-free adaptive trajectory tracking predictive control method and data-driven trajectory tracking predictive control method based on subspace identification.The proposed methods are based on dynamic linearization technique and subspace identification technique,respectively.Both of them use the measurable input and output data of the system to establish a data model,and further establish unconstrained and constrained data-driven trajectory tracking methods according to the 4D trajectory tracking object and predictive control concept.Numerical simulations verify the effectiveness of the proposed methods.4.Two model-free adaptive control scheme are designed to deal with the 4D trajectory tracking and flight attitude control problem during unsteady flight,namely modelfree adaptive integrated control and decentralized estimation decentralized control.Both of the methods are based on the full format dynamic linearization technique and uses system input and output data within a certain length of sliding time window at the current moment to construct a dynamic linearization data model.Then an unconstrained data-driven integrated control method is derived according to the trajectory and attitude control object.Considering the input constraints,such as aircraft aileron deflection angle,elevator deflection angle,rudder deflection angle,thrust and climb gradient angle,the data-driven integrated and decentralized control methods with constraints are further designed.Numerical simulations verify the effectiveness of the proposed methods.