APPLICATION OF THE ANN AND RANDOM FOREST MODELS FOR PREDICTING THE AREA OF THE TUNNEL FACE AFTER THE BLASTING

Abstract

To evaluate the effectiveness of underground construction methods using drilling and blasting, the area of the tunnel face after blasting has been used for the following reason: it is the main determining factor in the amount of work needed when constructing tunnels. Therefore, determining and predicting the area of the tunnel face after blasting in advance will greatly affect the completion of underground works, construction progress, and safety. This paper utilizes ANN (Artificial Neural Network) and RF (Random Forest) algorithms to predict the area of the tunnel face after blasting based on four input parameters (input variables): average borehole depth on the tunnel face (l, m), area of the tunnel face after the blasting when designed (D, m²), specific charge used during the tunnel construction (q, kg/m³), and the rock mass rating (RMR) where the tunnel is located. Based on comparisons of the obtained results from these artificial intelligence models with the actual results from the construction process of the Deo Ca tunnel, Phu Yen, Vietnam (the RF model achieved R² values of 0.9058 and 0.891 for training and testing data, respectively. On the other hand, the ANN model had R² values of 0.8818 and 0.894. In the RF model, with RMSE values of 0.15975 and 0.1774 for training and testing data, respectively, with equivalent RMSE values of 0.1634 and 0.1683 for the ANN model), the paper confirms the ability to predict the area of the tunnel face after the blasting during construction with high accuracy using artificial neural networks ANN and Random Forest models.