Enhancing Rock Fragmentation in Mining: Leveraging Ensemble Classification Machine Learning Algorithms for Blast Toe Volume Assessment

The condition of the floor after an explosion is of utmost importance for safety, as it directly impacts its stability. Moreover, it exerts an influence on fragmentation, hence affecting subsequent processes. Monitoring and improving the state of the floor improve both safety measures and overall operating effectiveness. This research proposes an ensemble learning approach to accurately estimate the condition of the mine's toe floor. The suggested model integrates the Support Vector Machine (SVM), Random Forest (RF), Bagged Random Forest (BRF), and K-Nearest Neighbor (KNN) algorithms. The input factors for predicting the four classes of toe floor condition include rock strength and seven adjustable parameters. The suggested model's results are evaluated by comparing their prediction accuracy and Receiver Operating Characteristic. The suggested model enhances the forecasting precision achieved from the RF, BRF, KNN-Weighted kernel, KNN-Fine kernel, SVM by 90, 94, 82, and 91%, correspondingly. Therefore, the BRF model can be utilized as a dependable approach for predicting the condition of the floor after a blast, to improve the design of the blasting process. In addition, this work introduces a linear mathematical model that utilizes powder factor and burden distance to forecast toe blast.