Skip to Main content Skip to Navigation

Jumeau Hybride dans le cadre de systèmes complexes

Abstract : The benefits of a deep understanding of the technological and industrial processes of our world are unquestionable. Optimization, inverse analysis or simulation-based control are some of the procedures that can be carried out once the above knowledge is transformed into value for companies. This brings better technologies that end up greatly benefiting society. Think of a routine activity for many people today, such as taking a plane. All the above procedures are carried out in the plane design, on-board control and maintenance, culminating in a technologically resource-efficient product. This strong added value is what is driving Simulation Based Engineering Science (SBES) to make major improvements in these procedures, leading to noticeable breakthroughs in a wide variety of sectors (e.g. Healthcare, Telecommunications or Engineering, to cite only a few).However, SBES is currently confronting several difficulties to provide accurate results in complex industrial scenarios. One is the high computational cost associated with many industrial problems which severely limits or even disables the key processes described above. Another problem is that in other applications, the more accurate (and also more highly-time consuming) models are not able to take into account all the details that govern the physical system under study, with observed deviations that seem to escape our understanding.Therefore, in this context, novel numerical strategies and techniques are proposed throughout this manuscript to deal with the challenges that SBES is facing. To do that, different industrial scenarios are analyzedThe above panorama also brings a perfect opportunity to the so-called DynamicData Driven Application Systems (DDDAS), whose main objective is to merge classical simulation algorithms with data coming from experimental measures. This concept is envisaged thanks to the exhaustive development in Data Science.Within this scenario, data and simulations would no longer be uncoupled but rather they would form a symbiotic relationship which would achieve milestones inconceivable until these days. Indeed, data will no longer be understood as a static calibration of a given constitutive model but rather the model will be corrected dynamically as soon as experimental data and simulations tend to diverge.For this reason, the present dissertation placed a particular emphasis on Model Order Reduction (MOR) techniques, as they are not only a tool to reduce computational complexity, but also a key element in meeting the real time constraints arising from the DDDAS framework.Furthermore, this thesis presents new data-driven methodologies to enrich the so-called Hybrid Twin paradigm. A paradigm which is motivated because it makes DDDAS possible. How? by combining parametric solutions and the MOR framework with “on-the-fly” data-driven (i.e. machine learning) correction models.
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Tuesday, May 31, 2022 - 12:18:18 PM
Last modification on : Wednesday, September 28, 2022 - 5:50:15 AM


Version validated by the jury (STAR)


  • HAL Id : tel-03682852, version 1


Abel Sancarlos Gonzalez. Jumeau Hybride dans le cadre de systèmes complexes. Génie mécanique [physics.class-ph]. HESAM Université; Universidad de Zaragoza (Espagne), 2021. Français. ⟨NNT : 2021HESAE053⟩. ⟨tel-03682852⟩



Record views


Files downloads