Accurate Numerical Computation of Hot Deformation Behaviors by Integrating Finite Element Method with Artificial Neural Network

Accurate flow behaviors characterization plays a decisive role in the design and optimization of hot plastic forming processes with finite element method (FEM). In this study, the hot compression tests at elevated temperature were performed on a Gleeble 3500 thermo-physical test machine to acquire t...

Full description

Saved in:
Bibliographic Details
Published inInternational journal of precision engineering and manufacturing Vol. 19; no. 3; pp. 395 - 404
Main Authors Yu, Xiehua, Deng, Linmao, Zhang, Xiaoyun, Chen, Meilong, Kuang, Fengfei, Wang, Yuan
Format Journal Article
LanguageEnglish
Published Seoul Korean Society for Precision Engineering 01.03.2018
Springer Nature B.V
Subjects
Artificial neural networks
Back propagation algorithms
Back propagation networks
Compression tests
Computation
Computer simulation
Constitutive relationships
Deformation
Design optimization
Engineering
Finite element analysis
Finite element method
Forming
FORTRAN
High temperature
Hot pressing
Industrial and Production Engineering
Materials Science
Mathematical analysis
Neural networks
Numerical analysis
Regular Paper
Strain
Stress propagation
Stress-strain relationships
Superalloys
True stress
Well construction
Flow behaviors
Superalloy
Artificial neural network
Numerical computation
Online AccessGet full text

Cover

More Information
Summary:Accurate flow behaviors characterization plays a decisive role in the design and optimization of hot plastic forming processes with finite element method (FEM). In this study, the hot compression tests at elevated temperature were performed on a Gleeble 3500 thermo-physical test machine to acquire the true stress-strain data of GH4169 superalloy. Subsequently, an artificial neural network (ANN) with back-propagation algorithm was employed to learn the experimental true stress-strain data and then predict the constitutive relationships outside experimental conditions. It was revealed that the ANN can precisely track and predict the flow behaviors of GH4169 superalloy. The optimally-constructed and well-trained ANN model was written into a general finite element (FE) software platform via a user defined material subroutine programmed in Fortran language. Finally, the simulated hot compression tests with the FE model implanted ANN model were conducted. The intercomparisons between the experimental and simulated stroke-load curves revealed that integration of FEM with ANN is a feasible approach to conduct quality numerical computation for the varied hot plastic forming processes.
ISSN:2234-7593
2005-4602
DOI:10.1007/s12541-018-0047-6