ISSN: 0090-3973
Page Count: 8

Neural Network Approach for Prediction of Wrinkling Limit in Square Metal Sheet Under Diagonal Tension
Di, S
Monash University,

Thomson, PF

Abstract

A method of predicting the onset of wrinkling in the Yoshida Buckling Test, devised to simulate the wrinkling behavior in press-forming of sheet metal, has been developed in the present work by using an artificial neural network. The influence of different network architectures, learning parameters, and material coefficients has been investigated. The neural network was trained using data obtained by finite element analysis. The effectiveness of a neural network as a tool for predicting wrinkling limits in sheet metal-forming is examined. It is found that the trained neural network is capable of covering a wide range of material properties and its prediction of nominal strain at the onset of wrinkling is in reasonable agreement with the analytical results.

Keywords:
sheet metal, forming processing, wrinkling, buckling, neural network, training

Paper ID: JTE11327J
DOI: 10.1520/JTE11327J
ASTM International is a member of CrossRef.