Application of machine learning in determining design parameters of dredging vessels
Type:
Presented during:
CEDA Dredging Days 2024
Authors:
J.C. Goeree and M. Alvarez Grima
Abstract
In recent years, the development and use of artificial intelligence (AI), especially the application of machine learning (ML) techniques (ML is a part of AI), has transformed the landscape of technology across different industries. Machine learning is used in driver assistance systems for (self-driving) vehicles. In financial services machine learning is used to detect fraud. Other applications include logistics, where machine learning is deployed for optimizing the supply chain by forecasting demand. The design of a vessel depends on many variables, such as the main dimensions or total installed power etc. These variables typically are the results of engineering methods, which have been refined over time to ensure good design solutions. However, this approach takes time to work out a design up to a certain detail. The use of machine learning can reduce the lead time especially in the (preliminary) design process of a ship. Where the vessel is properly dimensioned and optimized based on the predetermined requirements. The objective of this paper is to predict main characteristics of a dredging vessel by means of machine learning. The prediction is based on historical data from previously build Trailing Suction Hopper Dredgers. The approach is to perform an exploratory data analysis (EDA). This yields an understanding of the dataset, identifies patterns and gives correlations between the parameters, such as the main dimensions or the total installed power. Subsequently, using the (highest) correlated parameters as an input, a deep learning model is developed. This model can accurately predict the desired characteristics of a dredging vessel, for instance the sailing speed. By using a machine learning model, which is based on previously built vessels, desired design parameters can be predicted with a high accuracy. Typically, the highest correlated parameters are used as input, reducing lead time and enhancing the prediction of different main characteristics.
Keywords: Artificial Intelligence, power prediction, machine learning, ship design, main dimensions