https://www.mdu.se/

Digital transformation is rapidly changing the business landscape, compellingcompanies to invest in it due to its several benefits, such as reduced time-to-value, improved quality, and the ability to overcome hardware limitations. Aprominent technology enabling digital transformation is digital twin. A digitaltwin is a virtual representation of a physical asset, system, or process, usedfor monitoring and controlling purposes. Implementing digital twins requirestackling challenging engineering tasks such as establishing accurate digitalrepresentations of physical systems or ensuring live data synchronization. Suchchallenges are exacerbated by a lack of research on software architecturesthat meet required quality attributes, their actual implementation, and otherengineering tasks. Software architectures serve as a blueprint and foundationfor building software-intensive systems that meet specific requirements. Theyprovide a comprehensive overview of a system components, their properties,and relationships. Reference architectures, which are based on establishedbest practices in a given domain, help in reasoning about a class of softwarearchitectures and facilitate the engineering of software-intensive systems such asdigital twins. Given the above premises, the ultimate goal of this research is todefine a standard-based reference architecture for digital twins in manufacturingto improve interoperability. We use existing standards as common ground tofacilitate technology transfer between academia and industry. In addition, wefocus on interoperability as it is a crucial quality attribute affecting the functionalsuitability and performance efficiency of digital twins.

The digital twin concept has emerged as a pivotal tool in digital transformation, offering digital representations of physical components, systems, or processes to enable remote monitoring and control. The functional suitability of digital twins relies on interoperable subsystems that can seamlessly and effectively exchange data. Consequently, achieving interoperability in digital twin solutions is crucial. However, this remains an open challenge due to the diversity of data models employed across existing digital twin implementations. This challenge is further intensified by the lack of practical methods to integrate early-phase engineering models with digital twin models.This thesis provides software architecture analysis of digital twins, using the ISO 23247 standard as a baseline, and highlights the limitations of current interoperability solutions—drawing on findings from systematic studies, industry surveys, and expert interviews.In addition, it addresses the identified limitations by progressively developing a solution rooted in model-driven engineering, and the Asset Administration Shell standard, supporting the development of more scalable and standardized digital twins.The proposed approach aims to automate the integration process between engineering models and digital twin models, eliminating the need for manual creation of transformation rules. A full-scale implementation has been realized using model transformations to automatically generate Asset Administration Shell-compliant models from artifacts described in Systems Modeling Language version 2. The approach is developed and validated following an iterative, test-driven development methodology through the translation of a representative set of Systems Modeling Language version 2 models into Asset Administration Shell models, demonstrating its feasibility, completeness, and correctness.