Robust updating from experimental measurements in computational dynamics
Abstract
In general, deterministic computational models are used for updating a computational model using experiments. However, it is known that uncertainties have to be taken into account in order to improve the accuracy of the predictions, for instance in introducing a probabilistic model. Such an updating is called robust updating. Until now, most of the published works in this area concern the robust updating of dynamical systems in the low-frequency range with respect to data uncertainties using experiments, but model uncertainties are not taken into account. The present work proposes a methodology for robust updating of stochastic computational models in structural dynamics, for low- and mid-frequency ranges for which experiments are available. The stochastic computational model is constructed by the nonparametric probabilistic approach allowing model and data uncertainties to be taken into account. The cost function depends on the updating parameters made up of the mean parameters and the dispersion parameters of the probabilistic model.