Statistically-based estimation techniques are presented
in this study. These techniques incorporate structural test
data to improve finite element models used for dynamic
analysis.
Methods are developed to identify optimum values of the
parameters of finite element models describing structures.
The parameters which may be identified are : element area,
mass density, and moment of inertia; lumped mass and stiffness;
and the Rayleigh damping coefficients. A technique is
described for incorporating hydrodynamic effects on small
bodies by identifying equivalent structure mass, stiffness,
and damping properties. Procedures are presented for both
the free vibration problem and for forced response in the
time domain.
The equations for parameter identification are formulated
in terms of measured response, calculated response,
the prior estimate of the parameters, and a weighting
matrix. The form of the weighting matrix is presented for
three identification schemes : Least Squares, Weighted Least
Squares, and Bayesian. The weighting matrix is shown to be
a function of a sensitivity matrix relating structural
response to the parameters of the finite element model.
Sensitivities for the forced vibration problem are derived
from the Wilson Theta equations, and are presented for the
free vibration problem.
The algorithm used for parameter identification is
presented, and its implementation in a computer program is
described.
Numerical examples are included to demonstrate the
solution technique and the validity and capability of the
identification method. All three estimation schemes are
found to provide efficient and reliable parameter identification
for many modeling situations. / Graduation date: 1993
Identifer | oai:union.ndltd.org:ORGSU/oai:ir.library.oregonstate.edu:1957/36152 |
Date | 10 August 1992 |
Creators | Kruchoski, Brian L. (Brian Louis) |
Contributors | Leonard, John W. |
Source Sets | Oregon State University |
Language | en_US |
Detected Language | English |
Type | Thesis/Dissertation |
Page generated in 0.0101 seconds