Predicting rock mass deformation modulus by artificial intelligence approach based on dilatometer tests

Abstract

Accurately assessing the mechanical behavior of jointed rock mass is one of the most important requirements in the geotechnical and mining engineering projects, including site selection, design, and successful execution. The mechanical behavior of rock mass is significantly affected by the deformation modulus which can be influenced by several parameters. In this paper, a new radial basis function neural network (RBFNN) model was developed to predict deformation modulus based on dilatometer tests at the Bakhtiary dam site, Iran. The model inputs, mostly acquired from geotechnical bore holes, are overburden height (H), rock quality designation (RQD), unconfined compressive strength (UCS), bedding/joint inclination to core axis, joint roughness coefficient (JRC), and filling thickness of joints. High accuracy of prediction was examined by calculating indices such as the variance accounted for, root-mean-square error, mean absolute error, and the coefficient of efficiency. Sensitivity analysis has been conducted on the RBFNN results of Bakhtiary dam site. Based on the obtained results, UCS and RQD are the most effective parameters and inclination of rock joint/bedding to core axis is the least effective parameter in the deformation modulus of rock mass.