Prediction and optimization of efficient ship design particulars through advanced machine learning approaches

S.M. Rashidul Hasan, Md Shariful Islam, Zobair Ibn Awal, Khandakar Akhter Hossain,
Prediction and optimization of efficient ship design particulars through advanced machine learning approaches,
Ocean Engineering,
Volume 341, Part 2,
2025,
122572,
ISSN 0029-8018,
https://doi.org/10.1016/j.oceaneng.2025.122572.
(https://www.sciencedirect.com/science/article/pii/S0029801825022553)
Abstract: Ship design optimization based on machine learning is an emerging and rapidly evolving field within naval architecture and maritime engineering. It uses machine learning techniques to enhance various aspects of ship design, performance, and efficiency. This research analyses the efficacy of feedforward neural networks and random forests as machine learning techniques, utilizing metrics such as Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R-squared (R2) values to identify the most effective model. Based on the chosen machine learning method, key ship design parameters are evaluated to find an energy-efficient ship design solution. The study, based on an extensive dataset of 281 entries and a thorough evaluation of both models, highlights the Neural Network's exceptional performance in predicting ship parameters for energy-efficient design. This suggests its potential superiority over the Random Forest model in most cases. The study extends its impact through an optimization phase, employing MATLAB's ‘FMINCON’ algorithm to refine engine power values for an Energy Efficiency Design Index (EEDI) vessel. The optimized results highlight the Neural Network and Random Forest models' comparable effectiveness in achieving lower EEDI values, emphasizing the practicality of both models in enhancing energy efficiency. Overall, this research contributes valuable insights into machine learning-based ship design optimization, highlighting the Neural Network's prowess in predictive accuracy and its potential application for creating more sustainable and efficient vessels.
Keywords: Machine learning; Neural network; Random forest; Optimization; Ship principal particulars; fmincon; Energy-efficiency design index