Vu Minh CHAU will defend his PhD thesis Constitutive modelling of nonlinear isotropic elasticity using deep regression neural networks on Friday, 13 October 2023, 13h00-16h00, Campus Belval in room MNO.1020.
This dissertation embarks upon an investigation into the application of a machine learning paradigm within the domain of computational mechanics. The central objective of this research works pertains to the exposition of the constitutive response function, employing Deep Neural Network-based surrogate models to characterise non-linear compressible hyperelastic materials. In order to establish a framework that ensures consistent and physically meaningful material laws, this inquiry discerns and subsequently integrates six fundamental physical prerequisites specific to hyperelasticity within the framework of DNN-based models. The challenges encountered are effectively surmounted by means of formulating hyperelasticity in principal space and refining both the architectural configuration of the DNNs and their training process to accommodate these pivotal physical requisites.