Application of fuzzy inference system for prediction of rock fragmentation induced by blasting

Appropriate prediction of rock fragmentation is a vital task in the blasting operations of open pit mines. Rock fragmentation is affected by various parameters including blast pattern and rock characteristics, causing understanding the process difficult. As such, application of the robust techniques such as artificial intelligence can be utilized in this regard. In this paper, a predictive model was developed to predict rock fragmentation using fuzzy inference system (FIS) in Sarcheshmeh copper mine, Iran. For this purpose, blasting parameters including burden, spacing, hole diameter, Schmidt hammer rebound number, density of joint, powder factor, and stemming length were considered as model inputs to predict rock fragmentation (D80). In addition, by using the same data, a multiple equation was proposed with the help of multiple regression analysis (MRA). Results of coefficient of determination (R (2)) between predicted and measured rock fragmentation were computed as 0.922 and 0.738 for FIS and MRA models, respectively. Moreover, root mean square error (RMSE) and variance account for (VAF) FIS model were obtained as 2.423 and 92.195 %, respectively, while these values were achieved for MRA technique as 4.393 and 73.835 %, respectively. Comparison of the performance indices of the predictive models showed the superiority of the FIS model over the regression technique. Results of sensitivity analysis indicated that burden, spacing, and powder factor are the most influential parameters on rock fragmentation.