Machine Learning Models for Predicting Maritime Vessel Fuel Consumption in Offshore Environments

Fuel expenditure constitutes a substantial proportion of the operational costs of the vessel. Environmental factors, including meteorological conditions and oceanic currents, significantly impact fuel efficiency. Consequently, fuel costs often represent 50–60% of the total cost, which requires maritime companies to pursue methodologies and technologies that mitigate energy consumption. This research focused on developing a robust fuel consumption prediction model for vessels operating under a leading Crew Transfer Vessel (CTV) operator Njord offshore [24], using real-time data from engine sensors. To achieve this, machine learning techniques, including linear regression, random forest [19], gradient boosting, extreme gradient boosting (XGBoost), support vector regression (SVR), and multilayer perceptron (MLP), were used. The input variables encompassed the characteristics of the vessel and the prevailing meteorological conditions at the time of transit. The efficacy of the six algorithms was evaluated using metrics such as mean absolute error (MAE), mean squared error (MSE), root mean squared error (RMSE), and the coefficient of determination (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>). In particular, the gradient boosting algorithm demonstrated outstanding performance and the most suitability for the dataset, achieving a R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> value of approximately 94% in predicting fuel consumption for vessels operating within the specified offshore environment.