Reconfigurable Autopilot Design using Nonlinear Model Predictive Control: Application to High Performance and Autonomous Aircraft

Abstract

The work presented in this thesis examines several aspects of Nonlinear ModelPredictive Control (NMPC) that display and confirm its promising potentials as apowerful reconfigurable control scheme. The effects of significant nonlinearities andthe intrinsically unstable nature of high performance fighter aircraft, among otherchallenges, have been shown to be well handled in the NMPC framework. Thiswork illustrates how complex control and stability augmentation measures (whichare normally realized through ad hoc mode switching strategies) can be formulatedand implemented as NMPC objectives and constraints. Further suggestions onrobustness strategies for model/plant mismatch and compensation for couplingeffects which are not properly accounted for, have been presented and examined inthis work. Results on fault tolerance of NMPC are also presented and discussed inthis thesis. In this direction, NMPC has been shown to have unique inherent faultdetection capabilities due to its effective utilization of feedback and its internalmodel predictions. Different types of actuator/control surface failures, includingextreme cases of total actuator failure are examined as test cases for the NMPCreconfigurable fault tolerant control scheme developed in this work. The NMPCautopilots are designed for an F-16 fighter aircraft, and the implementation andsimulations were done using ACADO nonlinear optimization solver, interfaced withthe MATLAB/Simulink environment.