Output-only Nonlinear System Identification of Civil Structures using Sparse Measurement Data
This research represents another innovation in my Ph.D. work. I have originally proposed and developed an innovative framework for output-only nonlinear system identification of civil structures based on nonlinear finite element (FE) model updating, utilizing only the measured structural response to earthquake excitations.
The proposed framework provides a computationally feasible approach for health monitoring and damage identification of civil structures when either input seismic excitations are not measured or the measured seismic excitations are incomplete, and/or noisy. Grounded in Bayesian inference methodology, the proposed framework simultaneously estimates the FE model parameters and the input ground acceleration time histories using only the measured dynamic response of the structure. Two approaches were developed in this research effort to solve the augmented input and parameter estimation problem: (a) a recursive maximum likelihood (ML) estimation approach, which reduced to a recursive nonlinear optimization method, and (b) a stochastic filtering approach based on the recursive maximum (MAP) estimation method, which reduced to an extended Kalman filtering method. As a part of this research, I have extended the direct differentiation framework in OpenSees to compute the FE response sensitivities with respect to the input ground accelerations.