Predication of Backbreak caused by the blasting operations of an open pit mine using Grey Wolf Optimizer (GWO) and Random Forest (RF) algorithms

Abstract

Listen

Translate article

Abstract The main goal of open pit mines is to create adequate rock breakage whilst reducing adverse outcomes like flyrock, ground vibration, and back break. Of these, back break (BB) is a serious consequence of blasting in open pit mines, as it frequently diminishes economic advantages and has a negative impact on mines’ safety. As a result, accurate BB prediction is critical for design of mine blast and other production operations. In this study, grey wolf optimizer (GWO) and random forest (RF) algorithms were implemented to predict BB. 61 categories of collected data from A and B mines of Sangan iron ore complex, Iran, were considered. Seven effective parameters on BB, i.e., the ratio of row spacing to burden (S/B), blasthole length (L), specific drilling (SD), sub-drilling (U), specific charge (P), stemming (T), and average charge in each blast-hole (Q) and their corresponding BB values were measured. To implement the suggested methods, in the first stage, 48 data sets were utilized as training phase data and remaining data sets were considered as test phase data. Then, the coefficient of determination parameter (R2) was employed for the training and testing data to evaluate the efficiency of the suggested models. The precision of the GWO and RF algorithms was further evaluated in comparison to multiple linear regression (MLR) analysis. The coefficient of determination values for the GWO, RF, and MLR for the training phase were 0.922, 0.948, and 0.643 respectively, while for the testing phase were 0.959, 0.966, and 0.733 indicating that both of the artificial intelligence approaches, GWO and RF, are more efficient than the MLR. Also, the calculated values of VAF and RMSE indicators reveal the GWO and RF algorithms can accurately predict BB values. Finally, the sensitivity analysis performed on the input parameters showed that the average charge in each blasthole (Q) has the greatest impact and specific drilling (SD) has the least impact on BB.