Gurson-Tvergaard-Needleman parameters identification using artificial neural networks in sheet metal blanking

Abstract

This paper presents a method for the identification of large used coupled damage model parameters in sheet metal blanking. The existing finite element models describe easily the elastoplastic behaviour occurring during the sheet metal blanking. However, the description of the damage evolution is much more delicate to appreciate. The proposed method connects finite element method (FEM) with artificial neural networks (ANN) analysis in order to identify the values of the Gurson-Tvergaard-Needleman (GTN) parameters. Blanking tests are carried out to obtain the experimental material response under loading. A finite element model is used to compute the load displacement curve depending on the GTN parameters. Via a full design of experiments, a numerical database is built up, which is used for the ANN training. The identification of the damage properties is done by minimizing the error between an experimental load displacement curve and a predicted one by the ANN function. The identified damage law parameters are validated on the other experimental configurations of blanking tests.