Analysis of attention mechanisms for the prediction of ship fuel oil consumption

Abstract

Carbon Dioxide (CO2) remains the dominant contributor to climate change in shipping with Heavy Fuel Oil (HFO) prevailing as the most significant fuel utilised in maritime transportation globally. Thus, while several technologies, including the consideration of renewable energies and alternative fuels, are being explored to contribute towards the Net Zero goal, the consumption of Fuel Oil (FO) continues to be of a substantial concern. Moreover, the optimal use of FO can lead to minimising CO2 emissions as well. This necessitates the development of more sophisticated tools to optimise onboard consumption, thereby facilitating the reduction of emissions and the associated operational costs. Accordingly, this paper analyses the use of an attention mechanism-based deep learning model for the prediction of FO consumption. A case study on a tanker vessel is conducted to assess the performance of this type of model, aiming to develop a decision-making tool for optimising ship FO consumption.