Welcome to the new version of CaltechAUTHORS. Login is currently restricted to library staff. If you notice any issues, please email coda@library.caltech.edu
Published January 1, 1996 | public
Report Open

Statistical system identification and applications to seismic response of structures

Yang, Chi-Ming

Abstract

A pragmatic and versatile statistical system identification framework is presented and applied to seismic response records of structures. The framework is based on the interpretation of probability as a measure of plausibility and on Bayesian statistical inference. Various classical system identification techniques can be derived and viewed as the special cases of the framework. However, the framework can provide a more informative interpretation of the identified optimal model. When the number of sampled input and output data from structures is large, useful asymptotic approximations of the analytical results are available. These asymptotic approximations are incorporated into the framework by introducing the definitions of system identifiability and model identifiability. New asymptotic approximation results are derived for the system un-identifiable case. From the viewpoint of asymptotic approximations, the system identification problem is a non-trivial global optimization problem. Two generalized trajectory methods, the homotopy scheme and the relaxation scheme, are presented which can be combined to provide a very robust numerical procedure for global optimization. Both methods can also be applied to find the roots of a set of nonlinear algebraic equations. Structural model updating is useful because it can be applied to structural health monitoring and is also desirable since the theoretically based stiffness matrix of a structure can be improved by using the measured structural response data. However, no well-accepted solution to this difficult problem has emerged primarily because it is an ill-conditioned and non-unique inverse problem. A single-stage structural model updating approach using the least-squares prediction-error system identification method and a substructuring technique is proposed and applied to simulated and real structural response data.

Files

9602.pdf
Files (7.8 MB)
Name Size Download all
md5:472cfffc88b875982827cb62265ede63
7.8 MB Preview Download

Additional details

Created:
August 20, 2023
Modified:
October 24, 2023