Input-state-parameter estimation of structural systems from limited output information

Abstract

A successive Bayesian filtering framework for addressing the joint input-state-parameter estimation problem is proposed in this study. Following the notion of analytical, rather than hardware redundancy, the envisaged scheme, (i) adopts realistic assumptions on the sensor network capacity; and (ii) allows for a certain degree of uncertainty in the structural information available throughout the life-cycle of the monitored structure. This uncertainty is quantitatively expressed via a parameter vector of known functional relationship to the structural matrices. An observer is accordingly established, which recombines the dual and unscented Kalman filters. The former aims at tackling the unknown structural excitations, while the latter solves the state and parameter estimation problem via an augmented state-space. An extensive parametric study on simulated structural systems under different measurement setups, excitation types and structural properties demonstrates the method's effectiveness. © 2019 Elsevier Ltd