The Use of Machine-Learning Techniques in Material Constitutive Modelling for Metal Forming Processes

Accurate numerical simulations require constitutive models capable of providing precise material data. Several calibration methodologies have been developed to improve the accuracy models. Nevertheless, a model’s performance is always constrained by its mathematical formulation. Machine learning (ML) techniques, such as artificial neural networks (ANNs), potential overcome these limitations. use ML for modelling very recent and not fully explored. Difficulties related data requirements training are still open problems. This work explores discusses techniques regarding in metal plasticity, particularly contributing (i) parameter identification inverse methodology, (ii) model corrector, (iii) data-driven using empirical known concepts (iv) general implicit approach. These approaches discussed, examples given framework non-linear elastoplasticity. To conveniently train approaches, large amount concerning behaviour must be used. Therefore, non-homogeneous strain field complex path tests measured with digital image correlation (DIC) used that purpose.