A Novel Method to Accelerate the Solution of Compliance Using Deep Learning for Topology Optimization

Jiaxiang Luo; Yu Li; Weien Zhou; Xianqi Chen; Wen Yao

doi:10.1007/978-981-16-7381-8_111

A Novel Method to Accelerate the Solution of Compliance Using Deep Learning for Topology Optimization

Jiaxiang Luo
, Yu Li
, Weien Zhou
, Xianqi Chen
, Wen Yao

National University of Defense Technology
Academy of Military Medical Science China

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

1 Scopus citations

Abstract

There are many successful advances in using deep learning techniques to accelerate topology optimization design and solve the problem of high computation costs. However, the finite element method (FEM) is still needed to calculate the compliance for the structure output by the training task that uses convolutional neural networks to accelerate the calculation of optimal structures, which leads to more time spent on model training. Therefore, a deep learning model is proposed to accelerate the solution of compliance instead of FEM. To improve the accuracy of the model prediction, we prepare 50,000 samples which are generated by the solid isotropic material with penalization (SIMP) and build the developed model using the appropriate number of network layers. The results show that the error rate of the predicted compliance is less than 0.01 and the developed model is high-precision. In addition, the training efficiency using the developed model is 3 times higher than that using FEM during the model training.

Original language	English
Title of host publication	Advances in Mechanical Design - Proceedings of the 2021 International Conference on Mechanical Design, ICMD 2021
Editors	Jianrong Tan
Publisher	Springer Science and Business Media B.V.
Pages	1781-1792
Number of pages	12
ISBN (Print)	9789811673801
DOIs	https://doi.org/10.1007/978-981-16-7381-8_111
State	Published - 2022
Externally published	Yes
Event	International Conference on Mechanical Design, ICMD 2021 - Changsha, China Duration: 11 Aug 2021 → 13 Aug 2021

Publication series

Name	Mechanisms and Machine Science
Volume	111
ISSN (Print)	2211-0984
ISSN (Electronic)	2211-0992

Conference

Conference	International Conference on Mechanical Design, ICMD 2021
Country/Territory	China
City	Changsha
Period	11/08/21 → 13/08/21

Keywords

Artificial neural network
Deep learning
Finite element method
SIMP
Topology optimization

Access to Document

10.1007/978-981-16-7381-8_111

Cite this

Luo, J., Li, Y., Zhou, W., Chen, X., & Yao, W. (2022). A Novel Method to Accelerate the Solution of Compliance Using Deep Learning for Topology Optimization. In J. Tan (Ed.), Advances in Mechanical Design - Proceedings of the 2021 International Conference on Mechanical Design, ICMD 2021 (pp. 1781-1792). (Mechanisms and Machine Science; Vol. 111). Springer Science and Business Media B.V.. https://doi.org/10.1007/978-981-16-7381-8_111

Luo, Jiaxiang ; Li, Yu ; Zhou, Weien et al. / A Novel Method to Accelerate the Solution of Compliance Using Deep Learning for Topology Optimization. Advances in Mechanical Design - Proceedings of the 2021 International Conference on Mechanical Design, ICMD 2021. editor / Jianrong Tan. Springer Science and Business Media B.V., 2022. pp. 1781-1792 (Mechanisms and Machine Science).

@inproceedings{014d81026e044ebab6bff5bd7c4cff4c,

title = "A Novel Method to Accelerate the Solution of Compliance Using Deep Learning for Topology Optimization",

abstract = "There are many successful advances in using deep learning techniques to accelerate topology optimization design and solve the problem of high computation costs. However, the finite element method (FEM) is still needed to calculate the compliance for the structure output by the training task that uses convolutional neural networks to accelerate the calculation of optimal structures, which leads to more time spent on model training. Therefore, a deep learning model is proposed to accelerate the solution of compliance instead of FEM. To improve the accuracy of the model prediction, we prepare 50,000 samples which are generated by the solid isotropic material with penalization (SIMP) and build the developed model using the appropriate number of network layers. The results show that the error rate of the predicted compliance is less than 0.01 and the developed model is high-precision. In addition, the training efficiency using the developed model is 3 times higher than that using FEM during the model training.",

keywords = "Artificial neural network, Deep learning, Finite element method, SIMP, Topology optimization",

author = "Jiaxiang Luo and Yu Li and Weien Zhou and Xianqi Chen and Wen Yao",

note = "Publisher Copyright: {\textcopyright} 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.; International Conference on Mechanical Design, ICMD 2021 ; Conference date: 11-08-2021 Through 13-08-2021",

year = "2022",

doi = "10.1007/978-981-16-7381-8\_111",

language = "英语",

isbn = "9789811673801",

series = "Mechanisms and Machine Science",

publisher = "Springer Science and Business Media B.V.",

pages = "1781--1792",

editor = "Jianrong Tan",

booktitle = "Advances in Mechanical Design - Proceedings of the 2021 International Conference on Mechanical Design, ICMD 2021",

}

Luo, J, Li, Y, Zhou, W, Chen, X & Yao, W 2022, A Novel Method to Accelerate the Solution of Compliance Using Deep Learning for Topology Optimization. in J Tan (ed.), Advances in Mechanical Design - Proceedings of the 2021 International Conference on Mechanical Design, ICMD 2021. Mechanisms and Machine Science, vol. 111, Springer Science and Business Media B.V., pp. 1781-1792, International Conference on Mechanical Design, ICMD 2021, Changsha, China, 11/08/21. https://doi.org/10.1007/978-981-16-7381-8_111

A Novel Method to Accelerate the Solution of Compliance Using Deep Learning for Topology Optimization. / Luo, Jiaxiang; Li, Yu; Zhou, Weien et al.
Advances in Mechanical Design - Proceedings of the 2021 International Conference on Mechanical Design, ICMD 2021. ed. / Jianrong Tan. Springer Science and Business Media B.V., 2022. p. 1781-1792 (Mechanisms and Machine Science; Vol. 111).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - A Novel Method to Accelerate the Solution of Compliance Using Deep Learning for Topology Optimization

AU - Luo, Jiaxiang

AU - Li, Yu

AU - Zhou, Weien

AU - Chen, Xianqi

AU - Yao, Wen

PY - 2022

Y1 - 2022

N2 - There are many successful advances in using deep learning techniques to accelerate topology optimization design and solve the problem of high computation costs. However, the finite element method (FEM) is still needed to calculate the compliance for the structure output by the training task that uses convolutional neural networks to accelerate the calculation of optimal structures, which leads to more time spent on model training. Therefore, a deep learning model is proposed to accelerate the solution of compliance instead of FEM. To improve the accuracy of the model prediction, we prepare 50,000 samples which are generated by the solid isotropic material with penalization (SIMP) and build the developed model using the appropriate number of network layers. The results show that the error rate of the predicted compliance is less than 0.01 and the developed model is high-precision. In addition, the training efficiency using the developed model is 3 times higher than that using FEM during the model training.

AB - There are many successful advances in using deep learning techniques to accelerate topology optimization design and solve the problem of high computation costs. However, the finite element method (FEM) is still needed to calculate the compliance for the structure output by the training task that uses convolutional neural networks to accelerate the calculation of optimal structures, which leads to more time spent on model training. Therefore, a deep learning model is proposed to accelerate the solution of compliance instead of FEM. To improve the accuracy of the model prediction, we prepare 50,000 samples which are generated by the solid isotropic material with penalization (SIMP) and build the developed model using the appropriate number of network layers. The results show that the error rate of the predicted compliance is less than 0.01 and the developed model is high-precision. In addition, the training efficiency using the developed model is 3 times higher than that using FEM during the model training.

KW - Artificial neural network

KW - Deep learning

KW - Finite element method

KW - SIMP

KW - Topology optimization

UR - https://www.scopus.com/pages/publications/85127912407

U2 - 10.1007/978-981-16-7381-8_111

DO - 10.1007/978-981-16-7381-8_111

M3 - 会议稿件

AN - SCOPUS:85127912407

SN - 9789811673801

T3 - Mechanisms and Machine Science

SP - 1781

EP - 1792

BT - Advances in Mechanical Design - Proceedings of the 2021 International Conference on Mechanical Design, ICMD 2021

A2 - Tan, Jianrong

PB - Springer Science and Business Media B.V.

T2 - International Conference on Mechanical Design, ICMD 2021

Y2 - 11 August 2021 through 13 August 2021

ER -

Luo J, Li Y, Zhou W, Chen X, Yao W. A Novel Method to Accelerate the Solution of Compliance Using Deep Learning for Topology Optimization. In Tan J, editor, Advances in Mechanical Design - Proceedings of the 2021 International Conference on Mechanical Design, ICMD 2021. Springer Science and Business Media B.V. 2022. p. 1781-1792. (Mechanisms and Machine Science). doi: 10.1007/978-981-16-7381-8_111

A Novel Method to Accelerate the Solution of Compliance Using Deep Learning for Topology Optimization

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this