Systematic Review of Machine Learning applications in Marine Engineering

The application of machine learning techniques is an ever-growing trend in engineering fields - with marine engineering being far from an exception. The authors present a total of 91 papers selected from the top 25% journals in the “Engineering, Marine” subcategory – selecting papers that applied machine learning algorithms to novel problems in marine engineering. The results show that most researchers prefer the utilization of tree-based algorithms, which were most commonly used across papers (46 papers accounting for different tree-based methodologies), with the single algorithms that were most popular being the multilayer perceptron (and other feed-forward artificial neural networks of the same shape), used in 24 papers and support vector machines (used in 21 papers). The analysis of goals and applied methods indicates that the most common analysis in marine engineering is performed on numerical, tabular, data – followed by the time-series numerical data. The most covered topic present in the research is optimizing the fuel use of vessels (11 papers), followed by the application of machine learning in the modeling of vessel dynamics and the preliminary vessel design (10 papers each). Based on the existing trends, the field of machine learning application for marine engineering will only continue to grow.