Answers are grounded in this paper and cite their sources.
Rock fragmentation is a key determinant of blasting efficiency, excavation productivity, and downstream processing costs in mining and civil engineering. Achieving optimal fragmentation depends strongly on accurate burden ( B ) selection, yet B estimation remains challenging because of the complex interactions between rock-mass characteristics and blasting behavior. Previous studies have demonstrated the feasibility of using geomechanical parameters for B prediction; however, their integration with optimized ensemble learning, uncertainty quantification, and model interpretability remains insufficiently explored. To address this gap, this study develops a hybrid machine-learning framework for predicting B using rock blastability index ( BI ), rock quality designation ( RQD ), unconfined compressive strength ( UCS ), density ( D ), and cohesive strength ( C ). Extremely Randomized Trees (ERT) were optimized using four metaheuristic algorithms: Gravitational Search Algorithm (GSA), Grey Wolf Optimizer (GWO), Teaching-Learning-Based Optimization (TLBO), and Whale Optimization Algorithm (WOA). Among the developed models, ERT-GWO achieved the highest prediction accuracy and the lowest prediction uncertainty on the available testing dataset. SHapley Additive exPlanations (SHAP) identified C as the most influential parameter, followed by RQD and UCS , highlighting the importance of geomechanical properties for B estimation. The proposed framework demonstrates the potential of an accurate and interpretable decision-support approach for B estimation under the geological and operational conditions represented by the Mouteh dataset. However, its applicability to other geological and mining environments requires external validation using larger and geographically diverse datasets.
No takes yet. Share an insight, caveat, or question.
Hasanipanah et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: