Prediction of blast-induced flyrock by using neural-imperialist competitive method (Case Study: Sungun Copper Mine)

Abstract

        This research focuses on conducting studies that predict the distance of blast-induced flyrock, which is an undesirable environmental phenomenon in open-pit mines. While there are experimental methods available for predicting blast-induced flyrock, the complex process of assessing the distance of flyrock has reduced the efficiency of these approaches. This study employs artificial intelligence methods and statistical techniques to forecast the flyrock distance in the Sungun copper mine. Thus, an Artificial Neural Network (ANN-MLP) and a new hybrid model of Artificial Neural Network (ANN) optimized by the Imperialist Competitive Algorithm (ICA), known as (ICA-ANN), are used to predict the flyrock distance, considering crucial parameters such as the number of holes, hole spacing, burden, total charge, specific drilling, charge per hole and specific charge. The results showed that the Artificial Neural Network, with RMSE, MAE, and R² error values of 9.31 m, 7.10 m, and 0.81, respectively, was able to predict the flyrock distance well compared to the measured data in the test phase. However, the implementation of the imperialist competitive algorithm optimizer in the neural network enhanced the prediction of the flyrock distance, yielding RMSE, MAE, and R² values of 5.66 m, 4.60 m, and 0.89, respectively. Finally, by performing sensitivity analysis on the input parameters of the flyrock distance, it was determined that the amount of explosive consumption and the number of holes have the greatest impact on the blast-induced flyrock distance.