Approche probabiliste des incertitudes de modèles et de données pour la simulation numérique en mécanique

Christian Soize (1)

1 LaM - Laboratoire de Mécanique

This paper deals with a nonparametric probabilistic approach of model uncertainties and data uncertainties in computational dynamics for linear and nonlinear dynamical systems. First the concept of the nonparametric probabilistic approach of uncertainties due to modeling errors and system-parameter uncertainties are introduced. A short overview of this nonparametric approach based on the use of ensembles of random matrices constructed with the maximum entropy principle is given. A numerical example proving the capability of the nonparametric probabilistic approach to take into account modeling errors is presented. Finally, we present an exemple relative to the robustness of the numerical simulation model with respect to model and data uncertainties in dynamics of a spatial structure

Domaines: Vibrations [physics.class-ph]; Vibrations [physics.class-ph]; Acoustique [physics.class-ph]; Acoustique [physics.class-ph]; Probabilités [math.PR]

Christian Soize. Approche probabiliste des incertitudes de modèles et de données pour la simulation numérique en mécanique. 17ème Congrès Français de Mécanique 2005 pp. 1-5, Troyes, France, 29 Août - 2 Septembre, 2005., AFM, Aug 2005, Troyes, France. pp.Pages : 1-20. ⟨hal-00773338⟩