Please use this identifier to cite or link to this item: http://dx.doi.org/10.14279/depositonce-12508
For citation please use:
Main Title: Combined ANN-FEM approach for spatial-temporal structural response prediction: Method and experimental validation
Author(s): Drieschner, Martin
Wolf, Christoph
Seiffarth, Friedrich
Petryna, Yuri
Type: Preprint
URI: https://depositonce.tu-berlin.de/handle/11303/13732
http://dx.doi.org/10.14279/depositonce-12508
License: https://creativecommons.org/licenses/by/4.0/
Abstract: The prediction of system outcomes like strains or displacement fields in real technical systems is demanding due to the presence of unavoidable uncertainties. These uncertainties should be considered, for example by different uncertainty models either based on probabilistic, possibilistic or other approaches. In this contribution, a non-linear stability analysis of a three-dimensional carbon fiber reinforced plastic (CFRP) considering aleatory and epistemic uncertainties is conducted. For the realistic incorporation of the uncertainties in the finite element model, thickness variations and geometrical inaccuracies have been detected in advance by non-destructive testing on a real structure made of CFRP. Additionally, the material parameters have been defined as stochastic variables based on reference studies in the literature. If the underlying deterministic model itself is also time-consuming, it can be useful to surrogate the overall numerical simulation. Strains and displacement fields have been measured in a symmetric three-point bending test and compared to the numerical predictions produced by artificial neural networks (ANN). A sensitivity analysis is finally conducted which clarifies the strong dependence of the outcomes on the fiber volume content, the structural thicknesses and the stiffness in fiber direction.
Subject(s): aleatory uncertainty
epistemic uncertainty
artificial neural networks
ANN
carbon fiber reinforced plastic
CFRP
global stability failure
aleatorische Unschärfe
epistemische Unschärfe
künstliche neuronale Netzwerke
KNN
carbonfaserverstärkter Kunststoff
CFK
globales Stabilitätsversagen
Issue Date: 21-Oct-2021
Date Available: 22-Oct-2021
Language Code: en
DDC Class: 620 Ingenieurwissenschaften und zugeordnete Tätigkeiten
Sponsor/Funder: DFG, 273721697, SPP 1886: Polymorphe Unschärfemodellierungen für den numerischen Entwurf von Strukturen
DFG, 312928137, Mehrskalige Versagensanalyse unter polymorphen Unsicherheiten für den optimalen Entwurf von Rotorblättern
Series: Preprint-Reihe des Fachgebiets Statik und Dynamik, Technische Universität Berlin
Series Number: 2021-02
TU Affiliation(s): Fak. 6 Planen Bauen Umwelt » Inst. Bauingenieurwesen » FG Statik und Dynamik
Appears in Collections:Technische Universität Berlin » Publications

Files in This Item:

Item Export Bar

This item is licensed under a Creative Commons License Creative Commons