Traditional drill-and-blast operations in surface and underground mining rely on empirical formulas like the Kuz–Ram fragmentation model and Maurer’s drilling equations. These classical models often fail to account for complex, heterogeneous subterranean geology. Recent advancements in Machine Learning (ML), Deep Learning (DL), and Internet of Things (IoT) sensors offer a powerful alternative. Operators can now replace static trial-and-error techniques with dynamic computational frameworks. This transformation enables precise real-time drilling adjustments, accurate rock fragmentation forecasts, and comprehensive multi-objective cost optimization across the entire mine value chain.
Rock drilling efficiency depends heavily on controllable operational variables such as Weight on Bit (WOB), Revolutions Per Minute (RPM), and mud flow rate (Q). Physical equations struggle to capture non-linear downhole interactions under varying Unconfined Compressive Strength (UCS). Advanced artificial intelligence models overcome these physical limitations. Top-tier algorithms, including Exponential Gaussian Process Regression (GPR) and hybrid Transformer–LSTM architectures with TS-Mixer, predict Rate of Penetration (ROP) with exceptional accuracy, reaching R² values as high as 0.9991. Embedded platforms like Hexagon Drill Assist automate rig control to boost productivity by over 30%. Furthermore, drill-shaft IoT sensors like DataCloud’s Rhino process real-time vibrations to quantify rock density, joint spacing, and strength directly at the drill face.
Accurate prediction of rock fragmentation after blasting is essential for downstream processing. Modern machine learning frameworks replace outdated empirical formulas with high-precision predictive engines. Hybrid models combining Backpropagation Neural Networks (BPNN) and Extreme Gradient Boosting (XGBoost) with metaheuristic optimizers like the Zebra Optimization Algorithm (ZOA) accurately forecast mean fragment size (P50). Explainable AI tools like SHAP consistently reveal that specific charge, powder factor, and burden distance account for over 80% of model predictive power. In addition to these algorithms, high-resolution drone mapping and 3D digital twin platforms provide centimeter-level terrain analytics. Mines utilizing these integrated spatial insights achieve significant performance gains, such as a 5.5% reduction in fragment size distribution when adopting electronic initiation.
Blast outcomes directly influence overall mine profitability and operational safety. Single-objective approaches fail because lowering drilling costs often increases crushing energy demands later. Modern AI frameworks solve this trade-off by pairing predictive engines with evolutionary multi-objective algorithms like NSGA-II. This combination generates a Pareto optimal front, allowing site engineers to select blast designs that balance explosive expenditure against target fragment size. Beyond immediate cost savings, uniform rock breakage reduces wear on loading shovels and lowers energy consumption in crushing circuits. Furthermore, precise AI blast modeling minimizes environmental hazards, effectively controlling ground vibration, flyrock, air overpressure, and backbreak to protect mine personnel and infrastructure.
Artificial intelligence represents a fundamental shift in subterranean engineering, replacing rigid empirical rules with dynamic data-driven insights. Machine learning models elevate drilling productivity while predicting rock fragmentation with unprecedented precision. Furthermore, multi-objective optimization aligns drill-and-blast parameters with whole-of-mine economic goals and safety standards. Explainable AI frameworks like SHAP and LIME provide necessary transparency, building user trust among site personnel. The adoption of these intelligent systems allows mining operations to minimize operational expenses, protect workers, and achieve long-term environmental sustainability.
