Aircraft Longitudinal Automatic Landing Control Using Model Predictive Control

Abstract

In this report, continuous model predictive control using orthonormal functions has been used to control landing trajectory of the aircraft. The purpose of this thesis was to implement constraint control on unmanned aerial vehicle and get improved results as compared to PID technique. For the simulations in MATLAB, longitudinal state space model of reliance 0.46 trainer aircraft has been used because of its stable natural dynamics. The model has two inputs four states and four outputs. The model is linearized at speci c trim conditions. Model predictive control(MPC) is used for designing controller required in landing for unmanned aerial vehicle. MPC can be simulated in real time and online optimization at every step can be performed. State space model is used and selected outputs are tracked. Landing control requires more than one state to be controlled at the same time so MPC proved to be a good candidate for designing such multi input multi output(MIMO)system. In real life applications, constraints are always present in all dynamical systems. MPC gives the possibility of adding constraints on inputs, input derivatives, states and output of a system. This gives MPC a priority when designing constraints systems. Simulation done in MATLAB shows good performance of landing phase of the ight by simply tuning the parameters of the laguerre functions.Also controller handled constraint very well while slightly compromising on the output. But over all it showed improved performance when compared to PID controllers.