Open Access
Subscription or Fee Access
Time Domain Response Surface Model Updating for Nonlinear Structures
Abstract
Finite element (FE) model updating aims to modify the uncertain parameters of a FE model to improve the correlation between certain analytical response features and their experimental counterparts. One of the proposed methods to solve this optimization problem is to approximate the input-output relationship of deterministic FE models with polynomial functions and estimate the uncertain parameters of the model using these approximated surrogate models, called Response Surface (RS) models. In this study, low computational effort associated with RS modeling is used to extend nonlinear FE model updating through time domain data. For this purpose, with assumption of known input and using least square techniques, RS models are constructed at every time step of the analysis and minimization problem of parameter estimation is solved iteratively in length of the time history of responses of the system. This paper investigates sensitivity of such RS estimates to measurement noise and input frequency in FE model updating of nonlinear structures. Further application of this procedure is investigated in a case study of a steel frame with bilinear material model under seismic loading.