Topology optimization of 2D structures with nonlinearities using deep learning

被引:153
作者
Abueidda, Diab W. [1 ,2 ,3 ]
Koric, Seid [1 ,2 ]
Sobh, Nahil A. [1 ,3 ]
机构
[1] Univ Illinois, Natl Ctr Supercomp Applicat, Champaign, IL USA
[2] Univ Illinois, Dept Mech Sci & Engn, Champaign, IL USA
[3] Univ Illinois, Beckman Inst Sci & Technol, Champaign, IL 61820 USA
关键词
Adjoint sensitivity; Finite element analysis (FEA); Machine learning; Neo-Hookean materials; Stress constraint; NEURAL-NETWORK; DESIGN; SIMULATION; MITIGATION; FRAMEWORK; ALGORITHM; STIFFNESS;
D O I
10.1016/j.compstruc.2020.106283
中图分类号
TP39 [计算机的应用];
学科分类号
081203 ; 0835 ;
摘要
The field of optimal design of linear elastic structures has seen many exciting successes that resulted in new architected materials and structural designs. With the availability of cloud computing, including high-performance computing, machine learning, and simulation, searching for optimal nonlinear structures is now within reach. In this study, we develop convolutional neural network models to predict optimized designs for a given set of boundary conditions, loads, and optimization constraints. We have considered the case of materials with a linear elastic response with and without stress constraint. Also, we have considered the case of materials with a hyperelastic response, where material and geometric nonlinearities are involved. For the nonlinear elastic case, the neo-Hookean model is utilized. For this purpose, we generate datasets composed of the optimized designs paired with the corresponding boundary conditions, loads, and constraints, using a topology optimization framework to train and validate the neural network models. The developed models are capable of accurately predicting the optimized designs without requiring an iterative scheme and with negligible inference computational time. The suggested pipeline can be generalized to other nonlinear mechanics scenarios and design domains. Published by Elsevier Ltd.
引用
收藏
页数:14
相关论文
共 102 条
[1]   Prediction of aeroelastic response of bridge decks using artificial neural networks [J].
Abbas, Tajammal ;
Kavrakov, Igor ;
Morgenthal, Guido ;
Lahmer, Tom .
COMPUTERS & STRUCTURES, 2020, 231
[2]   Mechanical Response of 3D Printed Bending-Dominated Ligament-Based Triply Periodic Cellular Polymeric Solids [J].
Abou-Ali, Aliaa M. ;
Al-Ketan, Oraib ;
Rowshan, Reza ;
Abu Al-Rub, Rashid .
JOURNAL OF MATERIALS ENGINEERING AND PERFORMANCE, 2019, 28 (04) :2316-2326
[3]   Prediction and optimization of mechanical properties of composites using convolutional neural networks [J].
Abueidda, Diab W. ;
Almasri, Mohammad ;
Ammourah, Rami ;
Ravaioli, Umberto ;
Jasiuk, Iwona M. ;
Sobh, Nahil A. .
COMPOSITE STRUCTURES, 2019, 227
[4]   Shielding effectiveness and bandgaps of interpenetrating phase composites based on the Schwarz Primitive surface [J].
Abueidda, Diab W. ;
Karimi, Pouyan ;
Jin, Jian-Ming ;
Sobh, Nahil A. ;
Jasiuk, Iwona M. ;
Ostoja-Starzewski, Martin .
JOURNAL OF APPLIED PHYSICS, 2018, 124 (17)
[5]   Acoustic band gaps and elastic stiffness of PMMA cellular solids based on triply periodic minimal surfaces [J].
Abueidda, Diab W. ;
Jasiuk, Iwona ;
Sobh, Nahil A. .
MATERIALS & DESIGN, 2018, 145 :20-27
[6]   Perspective: Materials informatics and big data: Realization of the "fourth paradigm" of science in materials science [J].
Agrawal, Ankit ;
Choudhary, Alok .
APL MATERIALS, 2016, 4 (05)
[7]   Bi-material topology optimization for energy dissipation with inertia and material rate effects under finite deformations [J].
Alberdi, Ryan ;
Khandelwal, Kapil .
FINITE ELEMENTS IN ANALYSIS AND DESIGN, 2019, 164 :18-41
[8]   Structural optimization using sensitivity analysis and a level-set method [J].
Allaire, G ;
Jouve, F ;
Toader, AM .
JOURNAL OF COMPUTATIONAL PHYSICS, 2004, 194 (01) :363-393
[9]   Artificial Neural Network Methods for the Solution of Second Order Boundary Value Problems [J].
Anitescu, Cosmin ;
Atroshchenko, Elena ;
Alajlan, Naif ;
Rabczuk, Timon .
CMC-COMPUTERS MATERIALS & CONTINUA, 2019, 59 (01) :345-359
[10]  
[Anonymous], PROC CVPR IEEE