0% found this document useful (0 votes)
48 views26 pages

Bayesian Model Updating

This document discusses applying Bayesian model updating to a shear building structure. It motivates the use of Bayesian updating due to modeling errors introducing uncertainty. Bayesian updating frames it as a statistical inference problem to identify parameter distributions rather than single values. It outlines using Markov chain Monte Carlo simulation and the Metropolis-Hastings algorithm to solve the problem. Examples are given of applying it to identify stiffness, mass, damping and other parameters of single degree of freedom systems and multi-story shear buildings.

Uploaded by

Kanta Prajapat
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
48 views26 pages

Bayesian Model Updating

This document discusses applying Bayesian model updating to a shear building structure. It motivates the use of Bayesian updating due to modeling errors introducing uncertainty. Bayesian updating frames it as a statistical inference problem to identify parameter distributions rather than single values. It outlines using Markov chain Monte Carlo simulation and the Metropolis-Hastings algorithm to solve the problem. Examples are given of applying it to identify stiffness, mass, damping and other parameters of single degree of freedom systems and multi-story shear buildings.

Uploaded by

Kanta Prajapat
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 26

BAYESIAN MODEL UPDATING: APPLICATION TO A

SHEAR BUILDING
(2012-13 SEMESTER II)
Previous Semester Progress
 Bayesian approach

 Bayesian approach an example

 Monte Carlo simulation

 Markov chain

 Metropolis hasting algorithm

2
Outline:

3
Motivation
 Response prediction of a structure

 Mathematical model describing the relation between input and


output of structure

 Discrepancies between measured output and model output of


a structure

 Our aim is to get the best model that can best fit the
observed data

4
Motivation
 In deterministic approach a single best model can be identified
by minimizing some least-squares fit between the measured
structural response and model response

 Gives non unique solution for the inverse problem

 Problem is due to the error associated with the process of


construction the mathematical model of the structure which
leads to uncertain accuracy in the predicted response

5
Sources of modeling errors:
 Variations of material properties during manufacture

 Inexact modeling of material constitutive behavior

 Uncertainties introduced during the construction process

 Inexact modeling of the boundary conditions

 Due to un-modeled features such as neglected nonstructural


components

6
Motivation
 Therefore the model updating can be best tackled as a
statistical inference problem.

7
Model updating as statistical inference problem

Measured = +
Model output Error
Output

Deterministic
Random part
part

Can be modeled as some


function of system
parameters

8
Bayesian statistical framework

Posterior distribution Likelihood function Prior distribution

o The likelihood of a set of parameter values given some


observed outcomes is equal to the probability of those
observed outcomes given those parameter values

9
Model updating as statistical inference problem

Measured = +
Model output Error
Output

Deterministic
Random part
part

Can be modeled as some


function of system
parameters

10
Model updating in Bayesian domain

More favorable Likelihood function Prior distribution


posterior based on of system
distribution observations parameters

Future Prediction of
response

11
Some Applications
of Model Updating

12
Single Degree of
freedom System

13
Damping ratio as unknown Parameter
Actual value = 0.02

14
Time Period and damping as unknown
parameters

Actual value = 0.02

15
Time Period and damping as unknown
parameters

Actual value = 2 sec

16
Shear Building

17
Stiffness as unknown (Frequency domain)

Actual value = 22000 N/m

18
Stiffness, mass and damping of a 4 storey shear
building as unknown (Time domain)

Actual value = 0.02

19
Stiffness
Actual value = 22000 N/m

20
Mass
Actual value = 2.8 kg

21
Model with different storey stiffness and
damping as unknown

Actual value = 0.02

22
First storey stiffness
Actual value = 240000 N/m

23
Second storey stiffness
Actual value = 180000 N/m

24
Third storey stiffness
Actual value = 120000 N/m

25
Fourth storey stiffness
Actual value = 60000 N/m

26

You might also like