Fragmentación de Roca por Voladura en Minería a Cielo Abierto: Predicción con IA

Rock fragmentation by blasting is fundamental in the mining sector. This process seeks to optimize rock size for subsequent extraction, transport, and processing. Predicting this fragmentation becomes a crucial task for improving operational efficiency; however, the application of simplified formulas lacks precision and adaptability to lithological variations. This study proposes the use of artificial intelligence (AI) for the accurate estimation of the average diameter of igneous, sedimentary, and metamorphic rocks. Based on a set of 97 real fragmentation and blasting data from around the world, the Random Forest (RF), Support Vector Regressor (SVR), and Kernel Ridge Regression (KRR) algorithms were evaluated, along with the Kuz-Ram equation, widely used in the industry. The results indicate that AI models significantly outperform the conventional equation. RF offers the highest accuracy with MSE of 0.0017 and R2 of 95.38%. In contrast, SVR achieves values of 0.0064 and 83.13%, while KRR achieves 0.016 and 69.60%. The adequate performance of these algorithms has led to the development of an application that allows users to view drilling grids, set specific dimensions, and compare projections of average rock size. This tool facilitates informed decision-making, improving mining processes and promoting more reliable and sustainable results in different operational contexts.