Stochastic Model Updating: Perturbation, Interval Method and Bayesian Inference

Stochastic model updating methods are described, including probabilistic perturbation methods, interval techniques and Bayesian inference. Particular attention is paid to aleatory uncertainty such as variability in nominally identical test structures due to manufacturing tolerances. In such cases the updating parameter distributions are meaningful physically either as PDFs or as intervals. Stochastic model updating is an inverse problem, generally requiring multiple forward solutions. The use of meta models (response-surface mapping techniques and the Kriging predictor) as surrogates for the full FE model are explained. The procedure is illustrated in each case by experimental examples, including model updating of a structure with uncertain locations of two internal beams.