Assessment of Supervised Machine Learning Models for their suitability in Flyrock Type Prediction induced by Mine Blasting

Blasting is an essential operation in mining and construction for breaking hard rock, but it inherently risks the dangerous phenomenon of flyrock, which can cause injury to personnel and damage to nearby infrastructure. Consequently, the accurate prediction of flyrock type (e.g., fractured, mid-wall (MW), or fresh) during mine blasting is critical for safety in construction areas. This study addresses this need by implementing eight different Supervised Machine Learning (SML) models—including Decision Tree (DT), Random Forest (RF), K-Nearest Neighbor (KNN), and Support Vector Machine (SVM)—to precisely forecast flyrock type based on real-time blasting parameters and rock mass properties. The models were trained and evaluated using a dataset comprising 152 blasting events recorded across three open-pit granite mines in Johor, Malaysia. The performance of the constructed SML models was rigorously assessed using standard validation metrics such as accuracy, precision, recall, and F1 Score. The results demonstrate the superior predictive capabilities of the ensemble models, with the Decision Tree, Random Forest, and AdaBoost models all achieving 100% accuracy in flyrock type prediction with zero prediction loss, thus providing robust tools for enhancing safety management in quarry and construction operations.