Advanced

A one step prediction error approach to the identification of viscoplastic material models

Wall, Ola LU and Holst, Jan LU (2003) IUTAM Symposium on Field Analyses for Determination of Material Parameters - Experimental and Numerical Aspects In IUTAM Symposium on Field Analyses for Determination of Material Parameters - Experimental and Numerical Aspects 109. p.101-112
Abstract
A new approach to the problem of estimating parameter in material models is presented. The approach is based on a state space representation of the constitutive equations and one step predictions. The differences between one-step predictions and the corresponding measurements are used to design generic one-step prediction error estimators, and in particular, the maximum likelihood method is presented. The one-step predictions are computed through extended Kalman filtering. Consequences of using a time dependent model with least squares regression are analysed. It is shown that if the residuals are a sequence of stochastic variables, correlated with the regressors, the parameter estimates may be biased. A Monte Carlo study shows that the... (More)
A new approach to the problem of estimating parameter in material models is presented. The approach is based on a state space representation of the constitutive equations and one step predictions. The differences between one-step predictions and the corresponding measurements are used to design generic one-step prediction error estimators, and in particular, the maximum likelihood method is presented. The one-step predictions are computed through extended Kalman filtering. Consequences of using a time dependent model with least squares regression are analysed. It is shown that if the residuals are a sequence of stochastic variables, correlated with the regressors, the parameter estimates may be biased. A Monte Carlo study shows that the model parameters of a Norton viscoplastic model are estimated with up to 40% higher precision with the new approach as compared to standard least squares regression. An analysis of the residuals clearly shows that the residuals of the new estimators form an independent sequence of random variables. (Less)
Please use this url to cite or link to this publication:
author
organization
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
keywords
identification, material models, visco-plasticity
in
IUTAM Symposium on Field Analyses for Determination of Material Parameters - Experimental and Numerical Aspects
volume
109
pages
101 - 112
publisher
Springer
conference name
IUTAM Symposium on Field Analyses for Determination of Material Parameters - Experimental and Numerical Aspects
external identifiers
  • wos:000183488300010
ISBN
978-1-4020-1283-9
language
English
LU publication?
yes
id
36a682aa-9b74-42a5-b8dd-ce5f3e7a17ce (old id 1406962)
date added to LUP
2009-06-03 13:51:20
date last changed
2016-04-16 07:42:42
@inproceedings{36a682aa-9b74-42a5-b8dd-ce5f3e7a17ce,
  abstract     = {A new approach to the problem of estimating parameter in material models is presented. The approach is based on a state space representation of the constitutive equations and one step predictions. The differences between one-step predictions and the corresponding measurements are used to design generic one-step prediction error estimators, and in particular, the maximum likelihood method is presented. The one-step predictions are computed through extended Kalman filtering. Consequences of using a time dependent model with least squares regression are analysed. It is shown that if the residuals are a sequence of stochastic variables, correlated with the regressors, the parameter estimates may be biased. A Monte Carlo study shows that the model parameters of a Norton viscoplastic model are estimated with up to 40% higher precision with the new approach as compared to standard least squares regression. An analysis of the residuals clearly shows that the residuals of the new estimators form an independent sequence of random variables.},
  author       = {Wall, Ola and Holst, Jan},
  booktitle    = {IUTAM Symposium on Field Analyses for Determination of Material Parameters - Experimental and Numerical Aspects},
  isbn         = {978-1-4020-1283-9},
  keyword      = {identification,material models,visco-plasticity},
  language     = {eng},
  pages        = {101--112},
  publisher    = {Springer},
  title        = {A one step prediction error approach to the identification of viscoplastic material models},
  volume       = {109},
  year         = {2003},
}