Structural system identification under seismic excitation via Markov chain Monte Carlo method

Abstract

The Markov chain Monte Carlo (MCMC) method is applied to structural system identification under known seismic excitation in this study. In the method, unknown structural parameters are identified by sampling from a Bayesian posterior distribution. The unnormalized posterior probabilities of the parameters are calculated based on the state-space model using the Kalman filter (KF), from which the estimates of structural responses are also obtained. The performance of the method is first investigated numerically. When the model in the KF is the same as the true model, the identified parameters are precise and consistent. However, if model errors exist, the parameters are inconsistent under different excitations. For the system responses, the results present good accuracies. The method is further applied to a full-scale building experiment from the E-defense shaking table. Both 2D and 3D system models are considered. The estimated structural responses are shown to be accurate.