m. bascompta

This paper attempted to estimate the amount of flyrock in the Angoran mine in Zanjan province, Iran using the gene expression programming (GEP) predictive technique. The input data, including flyrock, mean depth of the hole, powder factor, stemming, explosive weight, number of holes, and booster were collected from the mine. Then, using GEP, a series of intelligent equations were proposed to predict flyrock distance. The best GEP equation was selected based on some well-established statistical indices in the next stage. The coefficient of determination for training and testing datasets of the GEP equation were 0.890 and 0.798, respectively. The model obtained from the GEP method was then optimized using teaching– learning-based optimization algorithm (TLBO). Based on the results, the correlation coefficient of training and testing data increased to 91% and 89%, which increased the accuracy of the Equation. This new intelligent equation could forecast flyrock resulting from mine blasting with a high level of accuracy. The capabilities of this intelligent technique could be further extended to the other blasting environmental issues.

In this research work, a comprehensive study is conducted to predict flyrock as a typical and undesirable phenomenon occurring during the blasting operation in open-pit mining. Despite the availability of several empirical methods for predicting the flyrock distance, the complexity of flyrock analysis has resulted in the low performance of these models. Therefore, the statistical and robust artificial intelligence techniques are applied for flyrock prediction in the Sungun copper mine in Iran. For this purpose, the linear multivariate regression (LMR), imperialist competitive algorithm (ICA), adaptive neuro-fuzzy inference system (ANFIS), and artificial neural network (ANN) methods are applied to predict flyrock with effective parameters including the blasthole diameter, stemming, burden, powder factor, and maximum charge per delay. According to the attained results, the ANN model with the structure of 5-8-1, Levenberg-Marquardt as the learning algorithm, and log-sigmoid (logsig) as the transfer functions are selected as the optimal network with the RMSE and R2 values of 5.04 m and 95.6% to predict flyrock, respectively. Also it can be concluded that the ICA technique has a relatively high capability in predicting flyrock, with the LMR and ANFIS models placed in the next. Finally, the sensitivity analysis reveal that the powder factor and blasthole diameters have the most importance on the flyrock distance in the present work.