Federation of data and models to define digital twins

Context

Digital twins are virtual representations of real-world products, systems, or processes, enabling simulation, integration, testing, monitoring, and maintenance. They play a pivotal role in optimizing complex systems across a wide range of domains, from industrial manufacturing and energy to environmental monitoring and healthcare.

The Engineering Digital Twin EDT program, funded by the France 2030 investment plan, is a national initiative aimed at advancing the foundations of digital twin engineering in France and Europe. By bringing together leading academic and industrial partners, EDT seeks to strengthen the bases for the design, use, and deployment of digital twins, addressing key open challenges in model hybridization, composability, development methodologies, digital coupling, and human–twin interaction.

The construction of digital twins applies to many fields: industry for process or product control, territories to aid decision-making, but also medicine and archaeology, for example. Many digital twins are currently being built, often on an ad hoc basis, i.e. with a short-term operational objective.

In the long term, digital twins should be able to exist for the entire lifetime of their reference system (a factory, a product, a territory, etc.). Digital twins must therefore be designed and built to be easily maintainable, scalable, and interoperable, based on existing information and to collect all future information.

To this end, we see the digital twin as a repository of knowledge that must be modeled (in order to describe and use it) and built from highly heterogeneous sources of information. This heterogeneity has several sources: the nature of the information (observed values, 3D or behavioral models, physics or biology equations, documents, etc.) but also its temporality (real time, future, future in another configuration, past, past with other models). The aim of this thesis is to overcome this obstacle of heterogeneity.

Thesis Objectives

This PhD project aims to develop an information federation approach to design and maintain digital twins over the long term. Particular attention will be paid to the concept of a “timeline” to structure modeling and enable navigation in a multidimensional space of time, variants, and levels of granularity.

Two main questions arise:

• What organizational abstractions are necessary to ensure a long-term vision of digital twins, and

• How can model federation provide some answers?

To answer these questions, key scientific challenges include:

• Identify and classify models: descriptive—describes what is (to be maintained), prescriptive—describes what is interpreted as a constraint (to be validated or adjusted), predictive—describes what could be (to be validated)

• identify and classify descriptive data - source of truth (= reference), prescriptive - source of truth (= designer - to be validated or adjusted), predictive - source of truth (= predictor)

• describe the life cycles of these models and data in terms of their validity and traceability in order to navigate and compare developments in the past, present, and future of digital twins

• study ways of identifying descriptive elements (prescriptive, descriptive, or predictive) (data or model) and what they apply to; propose and define a vocabulary (ontology)

The results of this thesis will directly contribute to the Artemis platform, an open-source framework set to become a benchmark in the field.

Work Environment

The PhD candidate will be co-supervised by Antoine Beugnard, P4S/IMT-Atlantique and Jean-Marc Jézéquel, DiverSE/IRISA within LabSTICC/PS4.

The candidate will benefit from a stimulating scientific and industrial environment of the highest level, with access to a national network of leading research institutions and industry partners, regular interactions with the broader EDT community through workshops, seminars, and joint demonstrators, and the opportunity to contribute to Artemis, the program’s open software platform.

These two teams have been contributing for several years to research on digital twins, and in particular on the need to apply software engineering (SE) and model-based systems engineering (MBSE) principles to such systems.

It is part of the EDT program and will interact with other theses on related topics such as:

• TwinDevOps (PC3 TwinOps)

• Aggregation of digital twins as SoS (PC2 DTCompose)

• Variability (PC2 DTCompose)

What You Will Gain from This PhD

This PhD offers the opportunity to:

• Develop highly sought-after skills in system modeling, real-time data processing, and collaborative innovation.

• Collaborate with leading partners (Inria, CEA, CNRS, etc.) and validate your research on real-world industrial use cases.

• Join a network of PhD candidates within the EDT program, fostering collaboration, peer support, and interdisciplinary exchanges.

• Contribute to an open-source platform (Artemis) and publish in international conferences and journals.

• Gain recognition in a rapidly growing field, with career prospects in academic research, industrial R&D, or entrepreneurship.

Upon completion, you will be positioned as a recognized expert in a key domain for industry and research, with diverse professional opportunities in France and internationally.