Neural network constitutive modelling for non‐linear characterization of anisotropic materials

• DOI: 10.1002/nme.2999
• Publication Date: 2010-08-28T04:57:09Z
• Authors: H. Man; T. Furukawa
• Keywords: Characterization; Backpropagation

Abstract This paper presents a new technique of neural network constitutive modelling for non‐linear characterization anisotropic materials. The proposed technique, based on recently developed energy‐based framework, derives the variations external work applied to and strain energy induced in specimen. error between energies is subsequently correct properties by using modified backpropagation algorithm. Unlike conventional techniques modelling, develops models quantifying deformation specimen continuum basis. allows be constructed from single load test one Numerical examples first examine effect geometries loading conditions. noise experimental measurements investigated while having applicability behaviour shown thereafter. application materials finally demonstrated unidirectional lamina biaxial balanced laminate. Copyright © 2010 John Wiley & Sons, Ltd.