Abstract

En las jornadas Nacionales Españolas de Costas y Puertos se presenta las actividades de monitorización. En ella se expone que la operación de un buque está condicionada por su interacción continua con el medio físico, especialmente durante fases de alta exigencia como la aproximación al puerto, el acceso, el atraque, el amarre y la estancia atracado. En estas situaciones, la respuesta dinámica del buque constituye una fuente de información relevante para evaluar la seguridad, el confort, la eficiencia operativa y la exposición a condiciones meteo-oceánicas desfavorables. En el marco del proyecto ML-AMAR (Machine Learning Applications in Marine Engineering), se ha desarrollado una estrategia de monitorización continua orientada a integrar datos de navegación, respuesta dinámica y condiciones ambientales. Esta comunicación presenta su aplicación al buque Ro-Pax Ciudad de Barcelona, utilizado como demostrador en condiciones operativas reales. La metodología se basa en la instrumentación del buque mediante el sistema DeepMOTION-RTK, que permite registrar posición, velocidad, rumbo, aceleraciones y movimientos angulares mediante sensores GNSS/IMU. La campaña de monitorización cubre un periodo prolongado de operación, incluyendo navegación en ruta, aproximación, maniobras portuarias y estancia en puerto. Las señales registradas se integran con información meteo-oceánica procedente de fuentes externas, como Puertos del Estado y Copernicus, para relacionar la respuesta dinámica del buque con el oleaje, el viento y otras condiciones ambientales. Los resultados permiten estructurar la operación del buque por fases, identificar episodios de mayor solicitación dinámica y derivar indicadores trazables asociados a seguridad, confort y eficiencia. Asimismo, la información obtenida alimenta la plataforma DEEPVIEW, concebida para la visualización y análisis de los datos monitorizados, así como para la generación de métricas operativas de apoyo a la decisión. El trabajo confirma el interés de la monitorización dinámica como herramienta para mejorar la gestión operativa portuaria, validar modelos predictivos y avanzar hacia sistemas de apoyo basados en datos para la explotación de buques en condiciones reales. Palabras clave

Abstract

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 repository includes two public anonymized datasets generated from the ML-AMAR monitoring campaign. Both datasets have been prepared for scientific and technical distribution by removing sensitive information such as exact coordinates, precise timestamps, vessel/sensor identifiers and individual trajectories. The first dataset, ML_AMAR_anonimizacion_version_B_public, contains aggregated information on maritime routes, trip frequencies, departure time slots, trip duration classes, recurrent movement patterns and a non-georeferenced schematic network. It is intended for the analysis of maritime mobility and general route patterns. The second dataset, ML_AMAR_positions_1min_metocean_public_B, is derived from the original one-minute monitoring records. It provides aggregated information on operational state, route phase, distance-to-port classes, metocean conditions and vessel dynamic response. It is intended for scientific and technical analyses of the relationship between navigation, waves, wind and vessel behaviour. In both cases, the data are published only in aggregated and anonymized form. The repository does not include raw files, real GNSS positions, exact timestamps, operational identifiers or confidential correspondence tables between anonymized codes and real locations. Aggregation, discretization and suppression rules have been applied to prevent the reconstruction of individual trips or specific operational patterns. The dataset ML_AMAR_anonimizacion_version_B_public includes aggregated route summaries, trip frequencies, time slot distributions, duration classes, recurrent cycles and a schematic non-georeferenced network. Its main purpose is the analysis of maritime mobility and route patterns. The dataset ML_AMAR_positions_1min_metocean_public_B includes aggregated operational states, route phases, distance-to-port classes, vessel motion statistics and metocean conditions. Its main purpose is the scientific and technical analysis of vessel dynamics and metocean forcing.  

Abstract

During the design of a vessel or marine artefact, one of the main objectives is to ensure a long service life. The structure of these artefacts is essential to achieve this goal, as it is responsible for withstanding the extremely unfavourable stresses of the high seas. The aim of this paper is to present a method of detailed structural assessment by applying a dynamic hydroelastic model [1] that allows structural verification by greatly reducing the calculation time without losing accuracy. The current procedures used for this purpose require a detailed study of each loading condition, which implies many simulation hours and high computational cost. In this model, the high-fidelity structural solution is projected onto the modal basis to get the system modal matrix and extend the response amplitude operators (RAO) to the modal response amplitude operators (MRAO) of the structure. By retaining only those eigenmodes that preserve most of the structural elastic energy, the number of structural degrees of freedom can be significantly reduced. This reduced model has been implemented and coupled in the time domain with the seakeeping software SeaFEM, enabling quick analysis of a large number of load cases of the structure [2]. In this work, the first case study of a ship is presented.