Adaptive Control of Wind Assisted Propulsion Systems

As global efforts to mitigate climate change intensify, the maritime industry faces mounting pressure to reduce its greenhouse gas (GHG) emissions. Wind Assisted Propulsion (WAP) is a compelling solution, offering a direct and infrastructure-light method of supplementing conventional propulsion with renewable wind energy. While WAP systems promise significant reductions in fuel consumption and ancillary benefits such as decreased underwater noise pollution there is very limited research into optimising their control in operation. This paper investigates the use of Adaptive Control to optimise the performance of the AirWing, a compact WAP device ideal for use on vessels with limited deck space. A method for evaluating the performance of the control system separate from the device itself is first presented and then two model-less adaptive control methods, Extremum Seeking Control and Reinforcement Learning are developed and evaluated. Both are able to capture over 80% of the maximum available power saving with Reinforcement Learning capturing 96.2% suggesting significant potential for future real-world deployment.