Parameter identification using ROLSM in problems of fire technology

Jukka Myllymäki, Djebar Baroudi

Research output: Contribution to journalArticleScientificpeer-review

Abstract

This work presents a global reliable and systematic methodology regularised output least square method (ROLSM) to treat parameter identifcation problems. Several applications of the method to parameter identification in the problems of fire technology are presented. Numerical examples of heat transfer in insulated and bare metal structures are presented. For each problem one ordinary differential equation is derived from the variational formulation of the general heat conduction problem. As an example of the use of the method to mechanical properties of metals an identification of multiplicative viscoplasticity law for aluminium alloy AA46063-T6 is illus- trated. Since the inverse problem is usually ill-posed regularization is needed. Both regularization using mesh coarsing and Tikhonov-regularization (penalized output least square method) are used in order to get stabilized solution. The unknown distributed parameters are discretized using continuous piecewise linear basis functions.
Original languageEnglish
Pages (from-to)87-102
JournalRakenteiden Mekaniikka
Volume29
Issue number3-4
Publication statusPublished - 1996
MoE publication typeA1 Journal article-refereed

    Fingerprint

Cite this