Loading...

Automatic Landing in Cross Wind Using Model Predictive Control based on Differential Game

Majlesi Koopaei, Maryam | 2023

65 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 55966 (45)
  4. University: Sharif University of Technology
  5. Department: Aerospace Engineering
  6. Advisor(s): Nobahari, Hadi
  7. Abstract:
  8. This thesis proposes a robust model predictive controller for airplanes that can handle severe crosswind and indeterminate conditions, such as ground effect, without prior knowledge of wind direction and magnitude. The controller is a combination of extended dynamic model control (EDMC) model predictive control and differential game using a 6DOF nonlinear airplane model developed in MATLAB. The controller's cost function is designed to minimize controller effort during maximum disturbance and tracking error of desired outputs simultaneously. Testing for the landing phase is performed using a control structure in MATLAB for the longitudinal and lateral planes of the airplane. The structure includes controller loops for glide slope and flare phases in the ongitudinal plane, and crab and decrab phases in the lateral directional plane. The middle control loop in the longitudinal plane uses an H∞ controller, while the inner loop of the lateral directional plane control loop uses a controller based on multi-input multi-output EDMCDG. The overall structure of the longitudinal and lateral controllers is modified to fit the new controller. Simulations are conducted on a 6DOF nonlinear airplane model, and the controller's performance is evaluated in landing simulations with a 10 m/s crosswind, compared to the H∞ controller. Results demonstrate that the EDMCDG controller performs better in controlling lateral acceleration during touch down but not as well in the decrab phase. The controller prevents an increase in the lateral distance of the airplane from the landing band while the wind is continued, but cannot eliminate the steady-state error of the lateral distance from the landing band. Monte Carlo analysis is conducted on airplane landing simulations with the EDMCDG controller to further evaluate its performance. Additionally, risk analysis is performed on landing, evaluating the risk of decentered landings, landings with the steepest bank angle, and landings with the steepest sideslip
  9. Keywords:
  10. Predictive Control ; Automatic Landing ; Ground Effect ; Cross Wind ; Differential Game ; Extended Dynamic Matrix Control (EDMC)

 Digital Object List

 Bookmark

...see more