Ship speed prediction based on machine learning for efficient shipping operation

Bassam, Ameen M.; Phillips, Alexander B.; Turnock, Stephen R.; Wilson, Philip A.. 2022 Ship speed prediction based on machine learning for efficient shipping operation. Ocean Engineering, 245. 110449.

Before downloading, please read NORA policies.
1-s2.0-S0029801821017315-main.pdf - Published Version
Available under License Creative Commons Attribution Non-commercial No Derivatives 4.0.

Download (2MB) | Preview


Optimizing ship operational performance has generated considerable research interest recently to reduce fuel consumption and its associated cost and emissions. One of the key factors to optimize ship design and operation is an accurate prediction of ship speed due to its significant influence on the ship operational efficiency. Traditional methods of ship speed estimation include theoretical calculations, numerical modeling, simulation, or experimental work which can be expensive, time-consuming, have limitations and uncertainties, or it cannot be applied to ships under different operational conditions. Therefore, in this study, a data-driven machine learning approach is investigated for ship speed prediction through regression utilizing a high-quality publicly-accessible ship operational dataset of the ‘M/S Smyril’ ferry. Employed regression algorithms include linear regression, regression trees with different sizes, regression trees ensembles, Gaussian process regression, and support vector machines using different covariance functions implemented in MATLAB and compared in terms of speed prediction accuracy. A comprehensive data preprocessing pipeline of operational features selection, extraction, engineering and scaling is also proposed. Moreover, cross validation, sensitivity analyses, correlation analyses, and numerical simulations are performed. It has been demonstrated that the proposed approach can provide accurate prediction of ship speed under real operational conditions and help in optimizing ship operational parameters.

Item Type: Publication - Article
Digital Object Identifier (DOI):
ISSN: 00298018
Date made live: 25 Jan 2022 17:50 +0 (UTC)

Actions (login required)

View Item View Item

Document Downloads

Downloads for past 30 days

Downloads per month over past year

More statistics for this item...