The modelling of layered rocks using a numerical homogenisation technique and an artificial neural network

A method of creating a constitutive model layered rocks based on an artificialneural network (ANN) is reported in this work. The ANN gives implicitconstitutive function ∑ ∑n n1 F( E, ), relating the new state homogenizedstresses n 1 with old Σn and increment homogenized strains. first step to repeatedly run strain- controlled homogenisation onan uni-dimensional finite element periodic cell elastic-plasticmodels (Drucker-Prager) components. Paths are created ( , ) space,from which, set patterns formed train ANN. description how toprepare data discussion training issues presented. Finally,the procedure trained put into finite-element code (ZSoil.PC)as user-delivered function. approach verified by comparingthe results developed basing direct (single-scale)analysis, which showed acceptable accuracy.