Performance of ANN Model in Predicting the Bearing Capacity of Shallow Foundations

Abstract

This study deals with the performance of the artificial neural networks (ANNs) for predicting the ultimate bearing capacity of shallow foundations on cohesionless soil. From cone penetration tests (CPTs), footing dimensions and other soil parameters were considered as the input variables which have the most significant impact on bearing capacity predictions. The application of artificial neural network was carried out through the following steps; at first, we consider a total of 100 sets of data among which we used 89 sets of data for training to determine a relation between input variables and the bearing capacity of the soil. For testing and validation, other 11 data sets were used. The accuracy of the model was evaluated by comparing the results with conventional bearing capacity equations. Also, high coefficients of correlation, low root-mean-squared errors (RMSEs), and low mean absolute errors (MAE) were the indications to confirm that the ANN-based model predicts with much perfection.