Determination of constitutive properties from spherical indentation data using neural networks. Part I: the case of pure kinematic hardening in plasticity laws

In this paper the power of neural networks in identifying material parameters from data obtained by spherical indentation is demonstrated for an academic problem (pure kinematic hardening, given Young's modulus). To obtain a data basis for the training and validation of the neural network, numerous finite element simulations were carried out for various sets of material parameters. The constitutive model describing finite deformation plasticity is formulated with nonlinear kinematic hardening of Armstrong-Frederick type. It was shown by Huber and Tsakmakis (1998a) that the depth-load response of a cyclic indentation process, consisting of loading, unloading and reloading of the indenter displays a typical hysteresis loop for given material parameters. The inverse problem of leading the depth-load response back to the related parameters in the constitutive equations is solved using a neutral network. (C) 1999 Elsevier Science Ltd. All rights reserved.