Authors : Ye Si Thu Aung; Le Van Diem; Tran Hong Ha

Volume/Issue : Volume 11 - 2026, Issue 2 - February

Google Scholar : https://tinyurl.com/szvhm6mh

Scribd : https://tinyurl.com/cbwatk5c

DOI : https://doi.org/10.38124/ijisrt/26feb129

Note : A published paper may take 4-5 working days from the publication date to appear in PlumX Metrics, Semantic Scholar, and ResearchGate.

Abstract : Ship fuel consumption prediction and optimization is challenging due to limited operational data even often collected for many years for a ship, which is barely enough for training models, especially when limiting collecting only from ship logbook data. In this study, only 76 usable records were able to collected from 2 years of voyages from a bunker ship and it was trained with radient Boosting, Random Forest, and XGBoost, showed weak predictive performance, with R² values remaining below 0.64. To overcome this limitation, the training data were expanded using distribution-preserving augmentation, increasing the sample size to 1,000 while keeping the original statistical characteristics. After augmentation, prediction accuracy improved markedly, reaching an MAE of 0.027, an RMSE of 0.050, and an R² of 0.995. The improved Random Forest model was then used for fuel optimization. Four different optimization methods, Genetic Algorithm, RealCoded Genetic Algorithm, NSGA-II, and Particle Swarm Optimization were applied to adjust controllable variables such as vessel speed and trim under fixed voyage conditions. All four methods led to nearly the same optimal operating point and resulted in fuel savings of about 20.6 percent compared with the baseline voyage. This shows that once prediction stability is achieved, optimization results become consistent and reliable even when only logbook-scale data are available.

Keywords : Bunker Ship; Fuel Consumption Prediction; Limited Data; Machine Learning; Optimization.

References :

1. “2023 IMO Strategy on Reduction of GHG Emissions from Ships.” Accessed: Nov. 14, 2025. [Online]. Available: https://www.imo.org/en/ourwork/environment/pages/2023-imo-strategy-on-reduction-of-ghg-emissions-from-ships.aspx
2. United Nations Conference on Trade and Development, Review of Maritime Transport 2024. 2024. Accessed: Oct. 16, 2025. [Online]. Available: https://unctad.org/system/files/official-document/rmt2024_en.pdf
3. A. Fan, J. Yang, L. Yang, D. Wu, and N. Vladimir, “A review of ship fuel consumption models,” Ocean Engineering, vol. 264, p. 112405, Nov. 2022, doi: 10.1016/j.oceaneng.2022.112405.
4. R. Campbell, M. Terziev, T. Tezdogan, and A. Incecik, “Computational fluid dynamics predictions of draught and trim variations on ship resistance in confined waters,” Applied Ocean Research, vol. 126, p. 103301, Sep. 2022, doi: 10.1016/j.apor.2022.103301.
5. T. Uyanık, Ç. Karatuğ, and Y. Arslanoğlu, “Machine learning approach to ship fuel consumption: A case of container vessel,” Transportation Research Part D: Transport and Environment, vol. 84, p. 102389, Jul. 2020, doi: 10.1016/j.trd.2020.102389.
6. C. Papandreou and A. Ziakopoulos, “Predicting VLCC fuel consumption with machine learning using operationally available sensor data,” Ocean Engineering, vol. 243, p. 110321, Jan. 2022, doi: 10.1016/j.oceaneng.2021.110321.
7. S. Wang, B. Ji, J. Zhao, W. Liu, and T. Xu, “Predicting ship fuel consumption based on LASSO regression,” Transportation Research Part D: Transport and Environment, vol. 65, pp. 817–824, Dec. 2018, doi: 10.1016/j.trd.2017.09.014.
8. IN HEAVY VEHICLES,” International Journal of Innovative Research in Technology, vol. 9, no. 12, pp. 1230–1239, May 2023.
9. Y. Gu, Y. Wang, and J. Zhang, “Fleet deployment and speed optimization of container ships considering bunker fuel consumption heterogeneity,” MSE, vol. 1, no. 1, p. 3, Oct. 2022, doi: 10.1007/s44176-022-00003-2.