SHIP FUEL CONSUMPTION PREDICTION MODEL BASED ON SHIP LOGBOOK DATA: A CASE STUDYPages 22-29 Abstract
In this paper, an artificial intelligence (AI)–based model is proposed to accurately predict fuel consumption using operational data from ship logbooks. A ship logbook contains historical operational parameters such as ship speed, trim, draft, cargo load, ballast conditions, engine load, and weather conditions, which significantly impact energy usage. For training, two years of voyage data (2023–2024) from the bunker ship ASHICO VICTORIA were used. Fuel consumption is estimated as a function of these features. A total of 22 machine learning and deep learning algorithms were trained, tested, and compared, including Linear Regression, Ridge, SVR, HuberRegressor, Random Forest, Gradient Boosting, XGBoost, Artificial Neural Networks, and Deep Neural Networks. Performance was evaluated using MAE, RMSE, and R². Testing with raw data achieved MAE≈0.12, RMSE≈0.17, and R²≈0.93. This study supports energy management, operational optimization, and emission reduction in line with international regulations.
Keywords:
Artificial Intelligence (AI),
Energy Efficiency,
Fuel Consumption Prediction,
Ship Logbook Data.
|
Ye Si Thu Aung, 
PDF 