PREDICTION OF BLAST-INDUCED THE AREA OF THE TUNNEL FACE IN UNDERGROUND EXCAVATIONS USING FUZZY SET THEORY ANFIS AND ARTIFICIAL NEURAL NETWORK ANN

Abstract

The area of the tunnel face after the blasting determines the construction progress of the tunnel, the construction cost and the safety of the tunnel construction. Hence, it is a major concern to predict and subsequently control the area of the tunnel face after the blasting in tunnel excavations. This paper presented two artificial intelligence methods, the first method, the adaptive neuro-fuzzy inference system (ANFIS) and the second method, the artificial neural network (ANN) for the prediction of the area of the tunnel face after the blasting. 100 databases on blasting parameters and the area of the tunnel face after the blasting in practice at Deo Ca tunnel, Phu Yen, Viet Nam were used in this paper. On the basis of these data, models to predict the area of the tunnel face after the blasting using the ANN model and ANFIS model were built. The obtained results, including coefficient of determination (R²), mean squared error (MSE) of the ANFIS model (with values R²_training=0.9758; R²_testing=0.9705; MSE_training=0.009816; MSE_testing=0.014676). Besides, in the training and testing data sets, R² values of (0.9503, 0.9722) and MSE values of (0.0208, 0.0136) were found in the optimal ANN model. The results obtained by these proposed models were compared with the measured values. The results indicate that the proposed ANFIS and ANN models are applicable and accurate tools to predict the area of the tunnel face after the blasting with high accuracy.