Abstract
This article presents a comparative analysis of predictive models for bench blast fragmentation in underground mining operations, focusing on the performance of a novel multivariate regression model using cross-validation techniques against established surface blasting models. The study evaluates the efficacy of the multivariate regression model in predicting blast fragmentation from bench blasts in an underground metal mine, utilising separate datasets for model training and validation. The multivariate regression model is compared against two well-known and used fragmentation surface blasting models: the modified Kuz-Ram model and the Kuznetsov-Cunningham-Ouchterlony model. The performance evaluation metrics used include the mean absolute error, the mean absolute percentage error, the root mean square error and the relative root mean square error. Results demonstrate the effectiveness of the multivariate regression model in accurately predicting blast fragmentation in underground bench-blasting mining scenarios. The parameters in the model include burden, spacing, powder factor, rock factor, blastability index, stemming and relative weight strength to ANFO. Comparative analysis revealed superior performance in terms of predictive accuracy when compared to both the modified Kuz-Ram model and the Kuznetsov-Cunningham-Ouchterlony model. The multivariate regression model exhibited lower mean absolute error and root mean square error values, as well as reduced relative root mean square error values, indicating its capability to provide more precise predictions of blast fragmentation outcomes in underground environments. This research contributes to the advancement of bench blast fragmentation prediction techniques in underground mining, offering insights into the potential application of multivariate regression analysis in optimising blasting operations and enhancing productivity within underground metal mines.
Keywords
Get full access to this article
View all access options for this article.