| Summary: | The popularisation of sensors, internet of things, and data analysis tools is bringing opportunities for several industries to improve their products and processes. One of the concepts making use of such opportunities is the digital twin, a comprehensive simulation of an engineering system that receives updated data from that system to mirror and predict its behaviour. In the ship domain, development of comprehensive digital twins finds obstacles in the lack of interoperability among digital tools and their formats. This thesis investigates use of standards to model digital twin ships with the aim of enabling data exchanges among systems, i.e., a standards-based approach. The work begins by investigating challenges to digital twin standardisation from the perspective of previous initiatives in the ship industry, comparisons with other sectors, and emerging trends. A few drivers are identified to guide future standardisation attempts: pragmatism in scope, support to heterogeneous systems, openness, and intelligibility. Following from these drivers, the study identifies existing standards covering the domains of ship visualisation, sensor logs, and taxonomies for ship data. They are applied to develop a web-based digital twin of an experiment in a wave basin, taking advantage of the broad support offered to web standards. The case study delimits standardisation gaps in the domains of ship models, with its accompanying metadata, and simulation of ship behaviour in operation. Both of them are taken as motivation for following research stages. The first topic is addressed by extending the open source Vessel.js library, originally developed for conceptual ship design, to handle detailed vessel models suitable for digital twins. This is done with a flexible framework which allows mapping of digital twin data to existing ship taxonomies during both design and operation. The framework is applied to a case study with a research vessel, demonstrating advantages and uncovering obstacles to scaling. The second topic is ...
|
|---|