Ingénierie système dirigée par jumeau numérique : cadre formel et opérationnel
Auteur / Autrice : | Mama Diakite |
Direction : | Mamadou Kaba Traore |
Type : | Thèse de doctorat |
Discipline(s) : | Automatique, Productique, Signal et Image, Ingénierie cognitique |
Date : | Soutenance le 10/12/2024 |
Etablissement(s) : | Bordeaux |
Ecole(s) doctorale(s) : | École doctorale des sciences physiques et de l'ingénieur |
Partenaire(s) de recherche : | Laboratoire : Laboratoire de l'intégration du matériau au système (Talence, Gironde) |
Jury : | Président / Présidente : Yves Ducq |
Rapporteurs / Rapporteuses : Nabil Anwer, Franck Fontanili |
Mots clés
Résumé
Digitalization is now spreading to all sectors, leading to the emergence of an ''all-smart'' society. Intelligent systems are developing, widely integrating data and virtual technologies, thus becoming essential in our daily lives. In this context, the Digital Twin (DT) is emerging as a promising technology to manage the growing complexity of these systems by combining data-based and model-based engineering respectively. DT combines various technologies, including the Internet of Things, data science, Modeling & Simulation, and Artificial Intelligence, to monitor, understand, optimize and anticipate the perfor mance of a real system through its virtual version. Its use has rapidly spread to various sectors such as construction, manufacturing, aerospace and smart cities, playing a crucial role in their digital transformation. However, the deployment of DT solutions faces several obstacles. Among them, we note the lack of a consensual definition, the difficulties of synchronization between the DT and the physical system, as well as the absence of transversal engineering methods applicable to different domains, raising several research questions. This thesis addresses three main research questions : (1) how to unify the understanding of the DT concept ? (2) what strategies to adopt for the deployment of a DT ? and (3) what is the impact of synchronization on the quality of services provided by the DT. The objective is to develop a formal and operational framework for DT-driven system engineering, based on systems theory. The main contributions of this work are as follows : • A conceptual modeling approach to the DT : this is the development of a conceptual fra mework called DMS (Data-Model-Services) to capture the internal elements of the DT in its value chain. • A formal specification approach of the DT : it aims to eliminate ambiguities in the interpre tation of the concept, and to facilitate its symbolic manipulation. • An inference capability graph : it is a characterization of different levels of maturity of a DT in terms of its self-adaptation capabilities to changes in the real system to which it is paired, levels organized in a relational structure in the form of a graph. • A graphical modeling approach : the proposed graphical notation, which is based on sys tems theory and inspired by the DEVS (Discrete EVent System specification) paradigm, aims to simplify DT modeling efforts, reduce the risks of conceptual error, and promote the co development principle. • A DT architecture pattern : this is a generic model of modular and flexible software architec ture that captures in detail the interactions between the different components of a DT, in order to facilitate its software engineering, and to pave the way for the automatic or semi-automatic generation of executable DTs. • A conceptualization of the synchronization between a DT and its real peer : the characteri zation of the different factors specific to this synchronization aims to allow analyzing their impact on the quality of the results produced by the DT.•An urban DT prototype : this is the development of a prototype of a DT of the campus of the University of Bordeaux, in the context of a large-scale project, illustrating the practical implementation of the proposed framework. This work constitutes in our eyes a first step towards a DT-driven system engineering. Several challenges remain to be overcome in this direction, which are as many perspectives for our future research efforts, in particular the verification and validation of DT, the analysis and synthesis of model inference capabilities in the DT, the in-depth study of the sensitivity of the DT with regard to the synchronization options of the Real-DT couple, and design patterns for DT-driven system engineering.