Representation of material behavior: neural network-based models

Abstract

Material modeling has involved the development of mathematical models of material behavior from human observation of, and reasoning with, experimental data. Using a neural network to model material behavior is discussed as an alternative. The main benefits of using a neural network are that the behavior of a material can be represented within the unified environment of a neural network and that the neural network-based model is built directly from experimental data using the self-organizing capabilities of the neural network, i.e. the network is presented with the experimental data and learns the stress-strain relationships. The behavior of concrete in the state of plane stress under monotonic biaxial loading and under compressive uniaxial cyclic loading is modeled with backpropagation neural networks. The preliminary results from these neural network-based material models are satisfactory, and the approach shows promise in modeling the behavior of modern, complex materials, such as composites