Prediction of Groundwater Inflow Rate Using Non-Linear Multiple Regression and ANFIS Models: A Case Study of Amirkabir Tunnel in Iran

Abstract

The prediction of groundwater inflow rate (GIR) into a tunnel is one of the serious challenges during the design, construction and exploitation of tunnels. GIR could lead to undesirable effects on excavation process such as decrease in rock mass stability, make extra pressure on permanent and temporary stability system, destructive effects on geomechanical condition of rock and finally physical and economical dangers. In this paper, an adaptive neuro fuzzy inference system (ANFIS) has been presented for anticipating the GIR into AmirKabir tunnel, Iran. For this purpose, a total number of 110 datasets including most influential parameters on GIR were inquired and used to construct the GIR predictive model. To illustrate superiority of ANFIS model, a non-linear multiple regression (NLMR) model was also developed for anticipating of GIR. In order to assess the performance of the developed models, coefficient of determination (R2), root mean square error (RMSE) and variance account for (VAF) were calculated. The results of this research indicate the higher reliability of ANFIS compared to NLMR model for GIR prediction.