Efficiënte nonlinear model predictive control formulaties voor economische doelfuncties met luchtvaart- en automobielapplicaties

This thesis is concerned with optimal control techniques for optimal trajectory planning and real-time control and estimation. The framework of optimal control is a powerful tool which enjoys increasing popularity due to its applicability to a wide class of problems and its ability to deliver solutions to very complicated problems which cannot be intuitively solved. The downside of optimal control is the computational burden required to compute the optimal solution. Due to recent algorithmic developments and increases in the computational power, this burden has been significantly reduced over the last decades. In order to guarantee effectiveness and reliability of the solver, three main components are necessary: fast and robust algorithms, a good problem formulation, and a mathematical model tailored to optimisation. Indeed, both the model and the optimal control problem can usually be formulated in many different ways, some of which are better suited for optimisation. In this thesis we are concerned with all three components, with a focus on the last two. Concerning the problem formulation, we propose practical approaches for formulating optimal control, MPC and MHE problems in an optimisation- friendly fashion. Moreover, we analyse the stability properties of various MPC formulations, with a focus on so-called economic MPC, for which the stability theory is still developing. On the algorithmic level, we review the literature on optimisation and optimal control and we prove that it is possible to tune tracking MPC formulations in order to locally obtain the same behaviour as economic MPC. The main advantages of tuned tracking MPC over economic MPC consist in easier to guarantee closed-loop stability and applicability of efficient real-time algorithms. On the modelling side, we propose an approach for deriving models of reduced complexity and reduced nonlinearity for multibody mechanical systems. The use of nonminimal coordinates and DAE models enlarges the range of modelling possibilities and allows the control engineer to derive models which are better suited for optimisation. In order to provide an easy framework for the model derivation, we extend the Euler-Lagrange approach and we demonstrate how to implement the proposed approach in practice. In order to demonstrate the effectiveness of the proposed techniques, we deploy them for two applications: tethered airplanes and autonomous vehicles. Both examples are characterised by fast nonlinear constrained dynamics for which simple controllers cannot be deployed. Tethered airplanes are of particular interest because they are an emerging technology for wind energy production. In this thesis, we use optimal control to design trajectories which extract maximum energy from the airmass and compare single and dual-airfoil configurations. We moreover demonstrate the effectiveness of MPC and MHE for controlling the system in real time and apply the new tuning procedure for tracking MPC to show its ability to locally approximate economic MPC.nrpages: 274status: publishe