Structure evaluation using output-only systems identification methods and dynamic model updating

Abstract

Identifying the dynamic properties of a structure with precision is an
important point when trying to establish an adequate methodology
for monitoring special structures, since the occurrence of structural
damages modifies the original dynamic parameters. In this way,
too, it is also possible to calibrate and subsequently work with a
robust numerical model as a complement to evaluate structural
integrity. Currently, there is no recommended procedure in Brazil
to continuously monitor great structures. For this reason, this study
aims to work on the proposal of a continuous structural monitoring
system for the subsequent evaluation of structural “health”. Three
identification methods based solely on structural response (Peak
Picking, Reference-Based Data-Driven Stochastic Realization and
the Reference-Based Covariance-Driven Stochastic Realization)
are applied to the results of an experimental test on a threestory frame loaded with different excitation sources. Aspects
such as computational effort, precision and processing velocity
are analyzed. Subsequently, a method of model updating based
on the measured frequencies is also evaluated. The results show
that these methods can be an effective part of the intended
monitoring methodology.