A novel prediction method of fuel consumption for wing-diesel hybrid vessels based on feature construction

Accurate fuel consumption prediction is essential for optimizing the operation of wing-diesel hybrid vessels and improving energy efficiency. This paper proposes a grey box model (GBM) for wing-diesel hybrid vessel fuel consumption prediction based on feature construction. Both parallel and series grey box modelling methods, as well as six machine learning (ML) algorithms are adopted to establish twelve combinations of prediction models. Then, a feature construction method based on the aerodynamic performance of the wing and the energy relationship of the hybrid system is proposed. Three types of wing features, namely wing thrust, wing thrust power, and wing fuel consumption savings are constructed and introduced into each combination respectively. Finally, based on noon report data of a wing-diesel hybrid vessel, the combinations are trained and validated. The best combination is obtained by considering the root mean square error (RMSE), which is parallel modeling method, random forest (RF) algorithm, and wing fuel consumption savings feature. Its RMSE decreased by 41.7 % compared to the white box model (WBM). Therefore, the GBM proposed in this paper can predict the daily fuel consumption of wing-diesel hybrid vessels with high accuracy, facilitating operational optimization and contributing to the greenization and decarbonization of the shipping industry.