A comparative assessment of hierarchical control structures for spatiotemporally-varying systems, with application to airborne wind energy

Optimal control in a spatiotemporally varying environment is difficult, especially if the environment is partially observable. Altitude optimization of an airborne wind energy (AWE) system, in which the tower and foundation of a contemporary wind turbine is replaced by tethers and a lifting body, is a challenging problem of this kind. The wind velocity changes both spatially and temporally, and it can only be measured at the altitude where the system is flying, making the problem partially observable. In this work, we propose and evaluate hierarchical structures for the aforementioned problem, which fuse coarse, global for the chosen grid resolution and prediction horizon, where applicable control with fine, local control. These controllers leverage the advantages of both fine, local and coarse, global control schemes, while addressing their limitations. We show through simulation, using the real wind velocity data, that the hierarchical structures outperform legacy control strategies in terms of net energy generation.