ABSTRACT
The aim of model updating is to minimize the difference between the model prediction and the associated measurements by adjusting the model parameters. As stated in Brownjohn & Xia (2000) and Ren & Chen (2010), parameter selection is crucial step for model updating because the performance of model updating heavily depends on the chosen parameters to be calibrated. Generally, on the one hand, the number of updating parameters should be kept as small as possible, since the estimation of too many parameters may lead to an ill-conditioned numerical problem because of the limited number of measurements available. On the other hand, the selected parameters not only must be uncertain so that they need clarifying, but also should be sensitive to the target responses. It is a waste of resources to update a well-known parameter (i.e., with small uncertainty). Likewise, attempts to calibrate an insensitive parameter do not significantly improve the prediction performance of finite element model, but also likely cause the potential numerical ill-conditioning.
