Integrated Estimation/Identification Using Second-Order Dynamic Models

Abstract

An algorithm is presented which accurately identifies multi-input-multi-output systems characterized by vibrating structures. More specifically, an identification technique is integrated with an optimal estimator in order to develop an algorithm which is robust with respect to measurement and process noise. The unique functional form of the integrated approach utilizes systems described by second-order models. Therefore, theoretical mass, damping, and stiffness matrices, associated with lumped parameter models, are tailored with experimental time-domain data for system estimation and identification. This leads to an algorithm that is computationally efficient, producing realizations of complex multiple degree-of-freedom systems. The combined estimation/identification algorithm is used to identify the properties of an actual flexible truss from experimental data. Comparison of experimental frequency-domain data to the predicted model characteristics indicates that the integrated algorithm produces near-minimal realizations coupled with accurate modal properties.