Evaluation of effective geomechanical parameters in rock mass cavability using different intelligent techniques

The paper presents the results of a comprehensive investigation of the applicability of various intelligence methods for optimal prediction of rock mass caveability in block caving by using effective geomechanical parameters. However, due to the complexity of the prediction of rock mass cavability, artificial intelligence-based methods, including classification and regression tree (CART), support vector machines (SVM), and Artificial neural network (ANN), have been selected. For validating and comparing the results, common MVR was used. Because of the dependency of the modeling generality and accuracy on the number of data, we attempted to obtain an adequate database from the result of numerical modeling. The distinct element method (DEM) used to study the rock mass cavability. The results indicated that ANN is the most accurate modeling technique with a determination coefficient of 0.987 as compared with other aforesaid methods. Finally, the sensitivity analysis showed that joint spacing, friction angle, joint set number, and undercut depth are the most prevailing parameters of rock mass cavability. However, the joint dip has shown the minimum effect on rock mass cavability in block caving mining method.