Application of artificial intelligence techniques for identifying rock mass quality in an underground tunnel

Abstract

The horizontal probe drilling data from the Pahang-Selangor raw water transfer tunnel (PSRWT) was used to develop intelligence models that could be used to identify the rock mass quality ahead of the excavation. In this study, two common artificial intelligence techniques; artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS) were used to achieve this aim. The input variables include rock quality designation, discontinuity spacing, groundwater inflow, discontinuity conditions, and penetration rate. The target variable was the rock mass rating index. Coefficient of determination (R2), root mean squared error (RMSE), and variance accounted for (VAF) were calculated to assess the developed models. The R2, VAF, and RMSE values of 0.922, 92.08%, and 2.284 respectively for ANN indicates a lower prediction output than the ANFIS model with R2, VAF, and RMSE values of 0.925, 92.27%, and 2.054 respectively. The results show that ANFIS outperformed ANN in predicting rock mass rating.