a b s t r a c tThis paper presents a systematic approach for comparing a numerical model to test results performed in situ on a structure with time-dependent behavior. A crucial issue for an engineer is to be able to assess the quality of models, based on a series of measurements. Dealing with in situ measurements as experimental reference for model updating involves two major difficulties: the excitations can be multiple and affected by large disturbances. The validation process we propose is based on the mechanical concept of Constitutive Relation Error (CRE) and aims at reducing the Lack Of Knowledge (LOK) attached to both the excitation forces and the parameters of the numerical model. The updated values can be computed inside confidence intervals that correspond to the lower contours of the CRE-based residual to minimize.The proposed method will be illustrated with a numerical example taken from the aerospace industry and applied for correcting a simple ARIANE 5 model by comparison with flight measurements.
To cite this version:Paul-Baptiste Rubio, François Louf, Ludovic Chamoin. Fast model updating coupling Bayesian inference and PGD model reduction. Computational Mechanics, Springer Verlag, 2018, 62 (6) The paper focuses on a coupled Bayesian-Proper Generalized Decomposition (PGD) approach for the realtime identification and updating of numerical models. The purpose is to use the most general case of Bayesian inference theory in order to address inverse problems and to deal with different sources of uncertainties (measurement and model errors, stochastic parameters). In order to do so with a reasonable CPU cost, the idea is to replace the direct model called for Monte-Carlo sampling by a PGD reduced model, and in some cases directly compute the probability density functions from the obtained analytical formulation. This procedure is first applied to a welding control example with the updating of a deterministic parameter. In the second application, the identification of a stochastic parameter is studied through a glued assembly example.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.