Airborne Wind Energy (AWE) is a new power technology that harvests wind energy at high altitudes using tethered wings. Studying the power potential of the system at a given location requires evaluating the local power production profile of the AWE system. As the optimal operational AWE system altitude depends on complex trade-offs, a commonly used technique is to formulate the power production computation as an Optimal Control Problem (OCP). In order to obtain an annual power production profile, this OCP has to be solved sequentially for the wind data for each time point. This can be computationally costly due to the highly nonlinear and complex AWE system model. This paper proposes a method how to reduce the computational effort when using an OCP for power computations of large-scale wind data. The method is based on homotopy-path-following strategies, which make use of the similarities between successively solved OCPs. Additionally, different machine learning regression models are evaluated to accurately predict the power production in the case of very large data sets. The methods are illustrated by computing a three-month power profile for an AWE drag-mode system. A significant reduction in computation time is observed, while maintaining good accuracy. (c) 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).