Computational aspects of RBFN meta-models for reliability problems

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21110%2F16%3A00302928" target="_blank" >RIV/68407700:21110/16:00302928 - isvavai.cz</a>

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Computational aspects of RBFN meta-models for reliability problems

Popis výsledku v původním jazyce

The reliability-based design optimization (RBDO) seeks such designs that are cheap but still reliable. Despite the growing performance of computers, evaluation of reliability measures is still computationally expensive especially for small failure probabilities. In simulation techniques often used for the computation of reliability, the largest part of the computational time is devoted to a repeated evaluation of a performance function represented by a finite element model (FEM). In our approach, the original FEM model is replaced by a meta-model that has a similar response but it is faster to evaluate. Several samples still have to be evaluated with the original model; these samples are used for the meta-model construction. Radial basis functions (RBF) interpolation meta-models are still computationally expensive since every construction sample increases a dimension of a linear system. This contribution focuses on a reduction of the computational effort spent on the RBF model evaluation during RBDO. In comparison to our previous versions, the resulting system of equation is factorized subsequently with the adaptive updating of the meta-model leading to substantial computational time reduction.

Název v anglickém jazyce

Computational aspects of RBFN meta-models for reliability problems

Popis výsledku anglicky

The reliability-based design optimization (RBDO) seeks such designs that are cheap but still reliable. Despite the growing performance of computers, evaluation of reliability measures is still computationally expensive especially for small failure probabilities. In simulation techniques often used for the computation of reliability, the largest part of the computational time is devoted to a repeated evaluation of a performance function represented by a finite element model (FEM). In our approach, the original FEM model is replaced by a meta-model that has a similar response but it is faster to evaluate. Several samples still have to be evaluated with the original model; these samples are used for the meta-model construction. Radial basis functions (RBF) interpolation meta-models are still computationally expensive since every construction sample increases a dimension of a linear system. This contribution focuses on a reduction of the computational effort spent on the RBF model evaluation during RBDO. In comparison to our previous versions, the resulting system of equation is factorized subsequently with the adaptive updating of the meta-model leading to substantial computational time reduction.

Druh

D - Stať ve sborníku

CEP obor

JM - Inženýrské stavitelství

OECD FORD obor

—

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2016

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

32nd conference with international participation Computational Mechanics 2016 - Book of Extended Abstracts

ISBN

978-80-261-0647-0

ISSN

—

e-ISSN

—

Počet stran výsledku

2

Strana od-do

69-70

Název nakladatele

University of West Bohemia

Místo vydání

Pilsen

Místo konání akce

Špičák

Datum konání akce

31. 10. 2016

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—