Enabling digital twins in the maritime sector through the lens of AI and industry 4.0
Dimitrios Kaklis, Iraklis Varlamis, George Giannakopoulos, Takis J. Varelas, Constantine D. Spyropoulos
Abstract
Sustainability and environmental compliance in ship operations is a prominent research topic as the waterborne sector is obliged to adopt ”green” mitigation strategies towards a low emissions operational blueprint. Fuel-Oil-Consumption (FOC) estimation, constitutes one of the key components in maritime transport information systems for efficiency and environmental compliance. This paper deals with FOC estimation in a more novel way than methods proposed in literature, by utilizing a reduced-sized feature set, which allows predicting vessel’s Main-Engine rotational speed (RPM). Furthermore, this work aims to place the deployment of such models in the broader context of a cutting-edge information system, to improve efficiency and regulatory adherence. Specifically, we integrate B-Splines in the context of two Deep Learning architectures and compare their performance against state-of-the-art regression techniques. Finally, we estimate FOC by combining velocity measurements and the predicted RPM with vessel-specific characteristics and illustrate the performance of our estimators against actual FOC data.