Response Estimation Of Multi-Degree-Of-Freedom Nonlinear Stochastic Structural Systems Through Metamodeling

The ever-growing reliance on probabilistic performance-based frameworks in assessing and designing structural systems is creating a need for efficient tools for propagating uncertainty through general nonlinear and dynamic structural systems. This research is focused on the development of metamodeling strategies for rapid response evaluation of a class of nonlinear multi-degree-of-freedom (MDOF) structural systems driven by stochastic excitations. In particular, the nonlinear autoregressive with exogenous input (NARX) model has been demonstrated to be versatile and effective in this respect. However, significant difficulties in NARX model calibration and execution have been encountered when directly applying this approach to practical MDOF systems with large numbers of degrees of freedoms. To overcome this limitation, a new metamodeling approach is proposed in this work through combining projection-based model order reduction with multi-input multioutput NARX models. The effectiveness and accuracy of the proposed approach are illustrated on a 40-story nonlinear steel-frame subject to stochastic earthquake excitation.