An effective approach based on reliability methods for high-dimensional Bayesian model updating of dynamical nonlinear structures

Abstract

Abstract Bayesian model updating represents a sound formulation to incorporate the unavoidable uncertainties arising in the system identification of infrastructure assets. However, the treatment of cases involving a relatively large number of model parameters remains an open issue, especially for dynamic nonlinear structural models. In this context, an effective implementation of subset simulation is considered within the framework of Bayesian model updating with structural reliability methods (BUS). For improved numerical efficiency, a substructure coupling technique for dynamic analysis is implemented to develop a reduced-order model strategy. To assess the capabilities of the proposed method, an application example that considers a three-dimensional bridge model equipped with nonlinear devices is presented.