Scope

The project ML-AMAR (Machine Learning Applications in Marine Engineering) aims to develop machine learning tools to optimize the lifecycle management of ships, from design and operation to maintenance and structural monitoring. The proposal is motivated by the increasing need to reduce fuel consumption and greenhouse gas emissions in maritime transport, aligned with the International Maritime Organization (IMO) strategy, which establishes ambitious targets for reducing carbon intensity in international shipping by 2030 and 2050.

The scope of the project is structured around three main coordinated research lines. The first focuses on the development of Machine Learning (ML) models for predicting ship hydrodynamic behavior, especially seakeeping performance and wave-added resistance. Traditional numerical simulations used for these analyses are computationally expensive and time consuming. ML-AMAR proposes the use of machine learning models trained with large databases generated from advanced hydrodynamic simulations, enabling fast and accurate predictions during the early stages of ship design. These tools will help optimize hull forms and reduce the energy consumption and pollutant emissions of vessels.

The second research line addresses maritime route optimization using ML-based methodologies and high-resolution weather forecasting models. The project aims to create a comprehensive routing optimization tool capable of incorporating uncertainties in weather predictions, ship performance in waves, added resistance, and propulsive efficiency. To achieve this, meteorological and oceanographic data will be combined with optimization algorithms and machine learning techniques. The expected outcome is a decision-support system able to identify safer, faster, and more energy-efficient routes, reducing operational costs, emissions, and structural fatigue throughout the vessel’s operational life.

The third research area focuses on predictive maintenance and real-time structural health monitoring through the development of reduced-order models and digital twins. The project will develop computationally efficient hydroelastic solvers capable of predicting the structural response of ships under realistic sea conditions. These tools will support fatigue assessment, predictive maintenance strategies, and lifecycle analysis while significantly reducing computational costs compared to conventional finite element simulations.

Overall, ML-AMAR seeks to integrate artificial intelligence, numerical simulation, and data-driven methodologies into a holistic lifecycle management framework for ships. Its main objective is to improve the efficiency, sustainability, safety, and economic competitiveness of the maritime industry while supporting the transition toward greener and smarter shipping technologies. 

The ML-AMAR project has been funded under the 2021 programme Generación de Conocimiento 2021 of the Agencia Estatal de Investigación (project ref. PID2021-126561OB.