A novel intelligent ELM-BBO technique for predicting distance of mine blasting-induced flyrock

Abstract

Blasting is an economical technique for rock breaking in hard rock excavation. One of its complex undesired environmental effects is flyrock, which may result in human injuries, fatalities and property damage. Because previously developed techniques for predicting flyrock are having less accuracy, this paper develops a new hybrid intelligent system of extreme learning machine (ELM) optimized by biogeography-based optimization (BBO) for prediction of flyrock distance resulting from blasting in a mine. In the BBO-ELM system, the role of BBO is to optimize the weights and biases of ELM. For comparison purposes, another hybrid model, i.e., particle swarm optimization (PSO)-ELM and a pre-developed ELM model were also applied and proposed. To do so, 262 datasets including burden to spacing ratio, hole diameter, powder factor, stemming, maximum charge per delay and hole depth as input variables and flyrock distance as system output were considered and used. Many models with different combinations of training and testing datasets have been constructed to identify the best predictive model in estimating flyrock. The results indicate capability of the newly developed BBO-ELM model for predicting flyrock distance. The coefficient of determination, coefficient of persistence and root mean square error values of (0.93, 0.93 and 21.51), (0.94, 0.95 and 18.84) and (0.79, 0.85 and 32.29) were obtained for testing datasets of PSO-ELM, BBO-ELM and ELM model, respectively, which reveal that the BBO-ELM is a powerful model for predicting flyrock induced by blasting. The developed BBO-ELM model can be introduced as a new, capable and applicable model for solving engineering problems.