Skip to main navigation Skip to main content
  • E-Submission

JKSPE : Journal of the Korean Society for Precision Engineering

OPEN ACCESS
ABOUT
BROWSE ARTICLES
EDITORIAL POLICIES
FOR CONTRIBUTORS
REGULAR

다단 딥 드로잉 공정 두께 및 변형률 예측을 위한 심층 신경망 개발

Development of Deep Neural Network for Predicting the Thickness and Strain of Multi-Stage Deep Drawing Process

Journal of the Korean Society for Precision Engineering 2020;37(11):835-842.
Published online: November 1, 2020

1 부산대학교 대학원 항공우주공학과

2 부산대학교 부품소재산학협력연구소

1 Department of Aerospace Engineering, Graduate School, Pusan National University

2 Industrial Liaison Innovation Center, Pusan National University

#E-mail: bskang@pusan.ac.kr, TEL: +82-51-510-2310
• Received: June 4, 2020   • Revised: July 15, 2020   • Accepted: August 13, 2020

Copyright © The Korean Society for Precision Engineering

This is an Open-Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

  • 10 Views
  • 0 Download
  • 1 Crossref
  • 1 Scopus
prev next

Citations

Citations to this article as recorded by  Crossref logo
  • Coupled Analysis and Metamodel-Based Optimal Design of Interior Permanent-Magnet Synchronous Motor Considering Multiphysical Characteristics
    Dae Han Kim, Yong Min You
    International Journal of Automotive Technology.2025; 26(6): 1563.     CrossRef

Download Citation

Download a citation file in RIS format that can be imported by all major citation management software, including EndNote, ProCite, RefWorks, and Reference Manager.

Format:

Include:

Development of Deep Neural Network for Predicting the Thickness and Strain of Multi-Stage Deep Drawing Process
J. Korean Soc. Precis. Eng.. 2020;37(11):835-842.   Published online November 1, 2020
Download Citation

Download a citation file in RIS format that can be imported by all major citation management software, including EndNote, ProCite, RefWorks, and Reference Manager.

Format:
Include:
Development of Deep Neural Network for Predicting the Thickness and Strain of Multi-Stage Deep Drawing Process
J. Korean Soc. Precis. Eng.. 2020;37(11):835-842.   Published online November 1, 2020
Close

Figure

  • 0
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
Development of Deep Neural Network for Predicting the Thickness and Strain of Multi-Stage Deep Drawing Process
Image Image Image Image Image Image Image Image
Fig. 1 Design and output parameters of deep drawing process
Fig. 2 Geometry of target cup (mm)
Fig. 3 Max, minimum case of finite element analysis outputs
Fig. 4 Histogram of 500 sample outputs
Fig. 5 Structure of artificial neural network
Fig. 6 Actual vs predicted plot of max effective strain
Fig. 7 Actual vs predicted plot of minimum thickness
Fig. 8 Actual vs predicted plot of mean thickness
Development of Deep Neural Network for Predicting the Thickness and Strain of Multi-Stage Deep Drawing Process

Drawing ratio of each stage

Drawing ratio ( Blank thickness/Punch diameter ) * 100
1.0-0.6 0.6-0.3 0.3-0.15
1 Stage 0.50-0.53 0.55-0.58 0.58-0.60
2 Stage 0.76-0.78 0.76-0.78 0.78-0.79
3 Stage 0.79-0.80 0.79-0.80 0.80-0.81
4 Stage 0.81-0.82 0.81-0.82 0.83-0.85
5 Stage 0.84-0.85 0.84-0.85 0.86-0.87

Material properties of STS3048 (Adapted from Ref. 8 on the basis of OA)

Young’s modulus, E [GPa] 196
Poisson’s ratio, v 0.3
Yield stress, s 255
K [MPa] 1,505
n 0.65
ε ¯ 0 0.06

Design space

Stage Variable Sign Lower bound Upper bound
1 Punch corner radius [mm] Rp1 2 8
Die corner radius [mm] Rd1 2 8
2 Punch corner radius [mm] Rp2 2 8
Die corner radius [mm] Rd2 2 8
3 Die corner radius [mm] Rd3 2 8
All Coefficient of friction of punch-blank μp 0.1 0.3
All Coefficient of friction of die-blank μd 0.01 0.2

Performance of max effective strain prediction

DNN RBF-SVR
RMSE 0.0280 0.0362
MAE 0.0235 0.0264
R 2 0.7478 0.5794

Performance of minimum thickness prediction

DNN RBF-SVR
RMSE 0.0033 0.0038
MAE 0.0024 0.0030
R 2 0.7780 0.7136

Performance of mean thickness prediction

DNN RBF-SVR
RMSE 0.0022 0.0026
MAE 0.0017 0.0020
R 2 0.7344 0.6285

Max relative errors and design variables

Max effective strain Min thickness Mean thickness
Max relative error [%] 3.36 3.33 1.89
Variable Rp1 5.31 5.57 6.19
Rd1 4.57 6.11 4.50
Rp2 4.18 5.95 4.26
Rd2 3.56 4.98 4.89
Rd3 4.36 7.23 6.71
μp 0.247 0.296 0.230
μd 0.150 0.196 0.134
Table 1 Drawing ratio of each stage
Table 2 Material properties of STS3048 (Adapted from Ref. 8 on the basis of OA)
Table 3 Design space
Table 4 Performance of max effective strain prediction
Table 5 Performance of minimum thickness prediction
Table 6 Performance of mean thickness prediction
Table 7 Max relative errors and design variables