Session: 06-12-01 Ship Hydromechanics I

Paper Number: 78433

78433 - Application of Machine Learning Algorithms for Predicting Added Resistance in Arbitrary Wave Headings of a Ship 

During the voyage, the ship is subject to added resistance due to its surrounding wave conditions, which may directly affect the reduction of speed-power performance. Therefore, the estimation of added resistance in waves has been of great interest to many researchers. There are various ways to estimate it, but especially for purposes such as the initial ship design stage in which the dimensions are not determined or the performance analysis of global fleets that is difficult to obtain detailed hull shape, it had to rely on semi-empirical methods. In this study, we propose a machine-learning model that predicts added resistance in arbitrary wave headings using basic design parameters. First, extensive model experimental data on added resistance of ships were acquired. This dataset consisted of 51 ships with different types and sizes, including various operating conditions. To build machine learning models, algorithms such as XGBoost, random forest, artificial neural network, support vector regression, and k-nearest neighbors were considered, and dimensionless variables representing the hull form and operating condition were used as inputs. Through nested cross-validation, the evaluation of the model for the test dataset and hyperparameter tuning were performed together, which enabled performance comparisons of various algorithms. Finally, XGBoost showed the best performance, and additional interpretation of the model was possible through explainable AI methods. The machine learning-based added wave resistance model showed improved performance in terms of accuracy compared to existing semi-empirical methods without requiring detailed input and had the advantage of being predictable in arbitrary wave headings. Therefore, it is believed that the proposed model can be widely used for the purpose of estimating the wave effects on a ship across the maritime industry, such as ship design, speed-power performance evaluation in sea trial, and fuel consumption and greenhouse gas prediction.

Presenting Author: Young-Rong Kim Department of Marine Technology, Norwegian University of Science and Technology

Authors:

Young-Rong Kim Department of Marine Technology, Norwegian University of Science and Technology
Sverre Steen Department of Marine Technology, Norwegian University of Science and Technology