Paul Daum

Monday, January 20, 2020, 11:30

Room 02-014, Georges-Koehler-Allee 103, Freiburg 79110, Germany

This thesis demonstrates that nonlinear optimal control techniques are a real-time feasible option for accurate and reliable control of fast, constrained airplanes. It comprehensively covers the whole control-engineering pipeline, starting with mathematical modelling of the experimental setup, continuing with the identification the model parameters, and finishing with the design, implementation and evaluation of a closed-loop control system.
The setup is controlled using a Nonlinear Model Predictive Controller, tracking a nominal target trajectory that is pre-computed offline in a periodic optimal control problem. A Nonlinear Moving Horizon Estimator simultaneously computes accurate estimates of the system states using only the output of an inertial measurement unit. In a real-world experiment, empirical data was collected. The experimental results show that the identified model is an accurate representation of the rotational startup setup and that the proposed control system is capable of closely tracking periodic references in real-time.
With estimation, control and target selection comprehensively merged and successfully tested, this thesis represents a milestone towards fully autonomous rotational startup procedures for airborne wind energy systems.