GENERALIZED HARDENING PLASTICITY APPROXIMATED VIA ANISOTROPIC ELASTICITY - A NEURAL-NETWORK APPROACH

The present paper deals with a method which replaces an anisotropic ha rdening plasticity problem through an appropriate sequence of anisotro pic elasticity problems. The developed method is a parameter identific ation method realized in an appropriate neural network environment via supervised and unsupervised learning algorithms. Numerical examples c oncerning the prediction and/or the correction of interpolated or extr apolated yield surfaces in accordance to given experimental data to th e elliptic paraboloid failure surface illustrate the method developed.