Artificial Neural Networks for Modeling Drained Monotonic Behaviors of Rockfill Materials

In this paper, the feasibility of developing and using artificial neural networks (ANNs) for modeling the monotonic behaviors of different angular, rounded rockfill materials is investigated. The database used for development of the ANNs models comprises a series of 82 large scale drained triaxial tests. The deviator stress-volumetric strain versus axial strain behaviors were firstly simulated by employing the ANNs. A feedback model using Multi-Layer Perceptrons (MLPs), for predicting drained behavior of rockfill materials was developed in MATLAB environment and the optimal ANNs architecture is obtained by a trial-and-error approach in accordance to error indexes and real data. Reasonable agreements between the simulated behaviors and the tests results were observed, indicating that the ANNs are capable of capturing the behavior of rockfill materials. The ability of ANNs to prediction of the Hardening-Soil constitutive Model (HSM) parameters, residual deviator stresses and the corresponding volumetric strain were also investigated. Moreover, the ANNs generalization capability was also used to check the effects of items not tested, such as dry density, grain size distributions and Los Angles abrasion.