Improving Reliability of Kite Systems for Wind-Assisted Ship Propulsion

Abstract

Wind-assisted ship propulsion (WASP) using kites is a promising technology for reducing fuel consumption and greenhouse gas emissions, but reliable operation is critical due to highly variable wind conditions. This paper proposes a data-driven methodology to anticipate rare critical events during kite operation, based on experimental data provided by Beyond the Sea company. To demonstrate the approach, we focus on critical load events by defining rare occurrences through an artificial threshold on the measured traction force. A logistic regression model is trained to predict these events sufficiently in advance using sensor measurements, including kite position, traction force, wind speed and wind direction relative to the vessel. Furthermore, we investigate the impact of wind measurements on the model performance, evaluating how their inclusion affects the prediction of these rare events. The proposed algorithm can be integrated into the control architecture to trigger avoidance maneuvers, providing the controller with enough time to react and improving operational safety under critical conditions.