THE ESTIMATION OF ROCK MASS DEFORMATION MODULUS USING REGRESSION AND ARTIFICIAL NEURAL NETWORKS ANALYSIS

Rock mass deformation modulus (EM) is an important input parameter in geomechanical problems. Field tests to determine this parameter are time consuming and expensive. In this paper, two methods have been developed to estimate EM. In the first method, using regression analysis, five empirical equations have been obtained relating EM and the rock mass rating (RMR), with the polynomial fitting having the best correlation coefficient. In the other method, using artificial neural network (ANN), a model has been obtained for estimating EM based on the radial basis function (RBF). Finally, both methods are applied to estimate EM of Karun IV dam. The obtained values are compared with the results of in-situ test. The comparisons have shown that the accuracy of the ANN method is better than of the regression analysis.