PGD Based Model Reduction: Towards Dynamic Data Driven Application Systems - l'unam - université nantes angers le mans Accéder directement au contenu
Communication Dans Un Congrès Année : 2011

PGD Based Model Reduction: Towards Dynamic Data Driven Application Systems

Résumé

Process optimization and control problems can be addressed through a broad variety of approaches. The most common consists in considering the system as a black box relating a set of input variables to some outputs. Well established approaches of robust control and identification have been developed to handle, to some extend, sources of variability in the physical system, in the control system or in the data acquisition system. However, many systems need to be controlled and optimized at very fine levels that can only be simulated through computationally expensive models. By expensive, we understand that the time needed to solve the model is incompatible with the requirements of the control algorithm. For example in material processing the appropriate level of description usually is that of the fine microstructural scales involved in the physical and chemical transformations occurring during the process. The corresponding thermo-chemo-mechanical models are solved today at a pace that can be orders of magnitude too slow with respect to process control requirements. Moreover, as the relevant simulation scale is often connected to the different sources of variability through complex non-linear couplings, one simply cannot take advantage of system identification techniques or a posteriori model reduction such as the Proper Orthogonal Decomposition. These techniques would indeed yield too crude approximations of the real physical system.
ENUMATH_Leygue_Chinesta.pdf (58.67 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-01008658 , version 1 (08-05-2018)

Identifiants

  • HAL Id : hal-01008658 , version 1

Citer

Adrien Leygue, Francisco Chinesta, Elías Cueto. PGD Based Model Reduction: Towards Dynamic Data Driven Application Systems. ENUMATH 2011 Conference, 2011, Leicester, United Kingdom. ⟨hal-01008658⟩
286 Consultations
20 Téléchargements

Partager

Gmail Facebook X LinkedIn More