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
SPECIAL

인공 신경망과 유전 알고리즘을 이용한 관통자 형상의 최적화 설계

Optimization Design of Penetrator Geometry Using Artificial Neural Network and Genetic Algorithm

Journal of the Korean Society for Precision Engineering 2020;37(6):429-436.
Published online: June 1, 2020

1 서울과학기술대학교 대학원 기계디자인금형공학과

2 국방과학연구소 제4기술연구본부

3 서울과학기술대학교 기계자동차공학과

4 서울과학기술대학교 기계시스템디자인공학과

1 Graduate School of Mechanical Design and Manufacturing Engineering, Seoul National University of Science and Technology

2 The 4th Research and Development Institute, Agency for Defense Development

3 Department of Mechanical and Automotive Engineering, Seoul National University of Science and Technology

4 Department of Mechanical system and Design Engineering, Seoul National University of Science and Technology

#E-mail: cwlee@seoultech.ac.kr, TEL: +82-2-970-6371
• Received: March 11, 2020   • Revised: April 9, 2020   • Accepted: April 17, 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.

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

Citations

Citations to this article as recorded by  Crossref logo
  • A Study on 3D Printing Conditions Prediction Model of Bone Plates Using Machine Learning
    Song Yeon Lee, Yong Jeong Huh
    Journal of the Korean Society for Precision Engineering.2022; 39(4): 291.     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:

Optimization Design of Penetrator Geometry Using Artificial Neural Network and Genetic Algorithm
J. Korean Soc. Precis. Eng.. 2020;37(6):429-436.   Published online June 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:
Optimization Design of Penetrator Geometry Using Artificial Neural Network and Genetic Algorithm
J. Korean Soc. Precis. Eng.. 2020;37(6):429-436.   Published online June 1, 2020
Close

Figure

  • 0
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 10
  • 11
  • 12
  • 13
Optimization Design of Penetrator Geometry Using Artificial Neural Network and Genetic Algorithm
Image Image Image Image Image Image Image Image Image Image Image Image Image Image
Fig. 1 Schematic of penetrator geometry
Fig. 2 Schematic of Latin hypercube sampling method
Fig. 3 Schematic of the finite cavity pressure method
Fig. 4 Simulation model of the penetrator
Fig. 5 Material properties of VAR4340 with respect to strain rate6 (Adapted from Ref. 6 on the basis of OA)
Fig. 6 FEA model of penetrator geometry
Fig. 7 Results of the optimized hyper-parameters
Fig. 8 Schematic of the final artificial neural network
Fig. 9 Learning results of artificial neural network in (a) Training data and (b) Validation data
Fig. 10 Flow chart of genetic algorithm
Fig. 11 Optimal geometry of penetrator
Fig. 12 Comparison of penetration depths with the optimum geometry in (a) 5 types and (b) Projectile geometries
Fig. 13 Comparison of velocities according to time
Fig. 14 Deformation of the projectile after the penetration (a) Example_3 and (b) Optimization result
Optimization Design of Penetrator Geometry Using Artificial Neural Network and Genetic Algorithm

Range for parameter combination of penetrator geometry

Parameter Minimum [mm] Maximum [mm]
r 1 1.0 3.5
r 2 1.0 3.5
h 1 1.0 3.5
h 2 1.0 3.5

Range for the optimization of hyper parameter

Hyper parameter Minimum Maximum
Number of nodes at each hidden layer 5 100
L2 regularization coefficient 0.001 0.0001
Learning rate 0.01 0.0001
Number of iterations 1000 100000
Number of hidden layers 1 5
Activation function Sigmoid, Tangent hyperbolic, ReLU

Results of the optimized hyper-parameter

Hyper parameter Value
Number of nodes at each hidden layer 12
L2 regularization coefficient 7.44E-05
Learning rate 7.48E-03
Number of iterations 14100
Number of hidden layers 3
Activation function Sigmoid + Tangent hyperbolic

Optimal parameters of penetrator geometry

Parameter Optimized result
r1 [mm] 2.13
r2 [mm] 3.50
h1 [mm] 3.50
h2 [mm] 1.00
Mass [g] 20.01

Penetration depth and mass results according to projectile geometries

Type 1 2 3 4 Opt.
Penetration depth
[mm]
- 39 -106 -142 -183 -206
Mass [g] 19.8 19.3 20.7 20.5 20.0
Table 1 Range for parameter combination of penetrator geometry
Table 2 Range for the optimization of hyper parameter
Table 3 Results of the optimized hyper-parameter
Table 4 Optimal parameters of penetrator geometry
Table 5 Penetration depth and mass results according to projectile geometries