A New Constitutive Model for Prediction of Saturated Sand Behavior Using Artificial Neural Network

The soil constitutive relation is one of the important issues in soil mechanics[1]. In the current geotechnical approach, the relationship between stresses and strains is represented by a series of mathem atical equations to describe the soil behavior based on a set of fundamental parameters The main problem with this parametric approximation is the high mathematical complexity involved especially when nonlinear effects have to be included andalso when different types of soils are considered.In this paper by using test data from triaxial shear test of saturated sand, including void ratio, axial strain,deviatoric stress, confining pressure and mean effective stress, an Artificial Neural Network(ANN) is trained and tested. Finally, the results of the networks are employed to simulate the behavior of saturated sand and to compare the stress path curves obtained from test data with those of ANN results.