Variability of Modal Parameter Estimations Using Two Different Output-Only System Identification Methods

Abstract

Abstract Accurate damage identification in a structure depends on the accurate estimation of modal parameters; therefore, an important step in the vibration-based structural health monitoring process is the estimation of modal parameters. The purpose of this study was to investigate systematically the variability of identified modal parameters of a steel model bridge with low damping using simple statistical measures. In order to achieve this objective, first the modal parameters of an actual steel bridge were estimated using two different and widely used output-only system identification methods, the natural excitation technique with an eigensystem realization algorithm and enhanced frequency domain decomposition. Then a calibrated finite element model of the bridge was developed using the estimated modal parameters. Using the simulated data from the calibrated model, the variability of modal parameter estimations due to variability of the spatial sensor density, measurement/sensor noise, and response data length input factors was investigated. The input factors chosen are significant for the performance of the output-only system identification methods used and therefore are deemed important for accurate damage detection.