Utilizing heuristic strategies for predicting the backbreak occurrences in open-pit mines, Gol Gohar Mine, Iran
Parviz Sorabi,Mohammad Ataei,Mohammad Reza Alimoradi Jazi,Hesam Dehghani,Jamshid Shakeri,Mohammad Hosein Habibi
DOI: https://doi.org/10.1007/s00500-023-09613-8
IF: 3.732
2024-02-10
Soft Computing
Abstract:Backbreak (BB) is a detrimental outcome of blasting activities in mineral extraction processes within mines. It involves the development of fractures and cracks at considerable distances behind the last row of blast pits, leading to reduced bench safety and increased operational costs. Given the multitude of factors influencing BB, various techniques have been developed to predict and optimize its occurrence. This particular study focused on analyzing 48 blasts in the tailings section of Gol Gohar Mine No. 1 to forecast BB using the whale optimization algorithm (WOA), multiverse optimizer (MVO), sine cosine algorithm (SCA), ant lion optimizer (ALO), and multivariate linear regression (MLR). Comparative analysis of the four BB prediction models revealed that the MVO algorithm yielded the most favorable outcomes, with the train data exhibiting parameter values of 0.9901, 0.2161, 0.1127, 98.8472, and 0.0180 for R 2 , RMSE, MSE, VAF, and MAPE, respectively, while the test data displayed values of 0.6357, 1.4955, 1.2003, 63.5472, and 0.1951 for the same parameters. In addition, the analysis specifically emphasized the substantial influence of spacing, burden, and GSI as the primary determinants of the backbreak phenomenon. In stark contrast, however, powder factor, delay time, and joint condition are identified as having negligible effects on backbreak.
computer science, artificial intelligence, interdisciplinary applications