Application of artificial neural network for constitutive modeling in finite element analysis

In this paper a neural network-based finite element model will be presented for modeling of geotechnical engineering problems. The methodology involves incorporation of neural network based constitutive model in a finite element program as a substitute to conventional constitutive material models. The development of the algorithm will be presented followed by application of the intelligent finite element method to a number of engineering problems. The case studies considered will involve linear elastic, nonlinear elastic and elastoplastic material behavior. The results obtained from the analyses using the intelligent finite element method will be compared with those obtained from the standard finite element method with different conventional constitutive models. The results show that artificial neural networks have good potential as a unified approach to constitutive modeling in finite element analysis. The merits and limitations of the neural network based constitutive models will be discussed in detail.