Please use this identifier to cite or link to this item:
Title: Evaluation of Deep Learning Networks for Predicting Truss Topology Optimization Results
Authors: Görgülüarslan, R.M.
Ateş, G.C.
Keywords: Artificial neural network
topology optimization
truss structure
Chemical activation
Deep learning
Learning systems
Network architecture
Shape optimization
Structural optimization
Activation functions
Learning network
Neural network architecture
Structural optimisations
Thickness value
Topology optimisation
Truss optimization
Truss structure
Truss topology optimization
Neural networks
Issue Date: 2022
Publisher: American Society of Mechanical Engineers (ASME)
Abstract: The applicability of artificial neural networks (ANNs) on the prediction of the structural optimization results of a truss structure is investigated. Two different ANN architectures are employed and the effect of using various optimizers and activation functions on their prediction performance is evaluated. Unlike the traditional machine learning network strategies where usually a physical response of the truss optimization (such as compliance, stress, etc.) is predicted, in this study, a new way of prediction is utilized for the truss-like structures; particularly predicting the optimized thickness values of the strut members by the ANNs. Thus, the input data used in these networks are the thickness values of the strut members at a certain initial iteration while the optimized thickness values are predicted as the outputs. A cantilever beam example is presented for the truss optimization to show the efficacy of the presented ANNs. The results indicate that using the thickness values at a certain initial iteration as inputs and final iteration thicknesses as outputs in ANNs are promising for the structural optimization prediction of the presented truss problem with the appropriate selection of the architecture, optimizer, activation function, and the input-output data formation. Copyright © 2022 by ASME.
Description: ASME 2022 International Mechanical Engineering Congress and Exposition, IMECE 2022 -- 30 October 2022 through 3 November 2022 -- 186577
ISBN: 9780791886656
Appears in Collections:Scopus İndeksli Yayınlar Koleksiyonu / Scopus Indexed Publications Collection

Show full item record

CORE Recommender

Page view(s)

checked on May 22, 2023

Google ScholarTM



Items in GCRIS Repository are protected by copyright, with all rights reserved, unless otherwise indicated.