Jalalian Mohammad Hossein, Bagherpour Raheb, Khoshouei Mehrbod
Department of Mining Engineering, Isfahan University of Technology, Isfahan 8415683111, Iran.
Sci Rep. 2025 Jul 14;15(1):25383. doi: 10.1038/s41598-025-09491-0.
The Mine-to-Mill (MTM) approach is crucial in mining due to the high energy consumption and costs of comminution processes like crushing and grinding, which account for over 50% of total energy use. Optimizing these processes, starting from blasting, enhances efficiency and profitability. Accurate rock mass characterization is key to blasting optimization, and Monitoring While Drilling (MWD) provides real-time geotechnical data foron-the-spot adjustments. Acoustic emission monitoring, a leading MWD technique combined with intelligent models, offers promising results in rock characterization. This study employed a Support Vector Machine (SVM) model to predict rock mass properties from drilling acoustic signals. The model demonstrated high accuracy, achieving R² values of 0.976 (training) and 0.808 (testing). The Mean Absolute Percentage Error (MAPE) was 4.36% and 29.52%, while the Root Mean Squared Error (RMSE) reached 0.0486 and 0.141, the Mean Absolute Error (MAE) was 0.021 and 0.103, and the Mean Squared Error (MSE) was 0.0024 and 0.0199 for training and testing, respectively. These results confirm the model's reliability in estimating rock characteristics. Integrating acoustic emission monitoring with advanced modeling can enhance MTM strategies, reducing energy consumption, operational costs, and environmental impact in mining.
由于破碎和研磨等粉碎过程能耗高且成本高，占总能源使用量的50%以上，因此矿到厂（MTM）方法在采矿中至关重要。从爆破开始优化这些过程可提高效率和盈利能力。准确的岩体特征描述是爆破优化的关键，随钻监测（MWD）可提供实时岩土数据以便进行现场调整。声发射监测是一种领先的MWD技术，与智能模型相结合，在岩体特征描述方面取得了有前景的成果。本研究采用支持向量机（SVM）模型从钻孔声信号预测岩体性质。该模型显示出高准确性，训练集的R²值为0.976，测试集的R²值为0.808。训练集和测试集的平均绝对百分比误差（MAPE）分别为4.36%和29.52%，均方根误差（RMSE）分别达到0.0486和0.141，平均绝对误差（MAE）分别为0.021和0.103，均方误差（MSE）分别为0.0024和0.0199。这些结果证实了该模型在估计岩石特性方面的可靠性。将声发射监测与先进建模相结合可以增强MTM策略，降低采矿中的能源消耗、运营成本和环境影响。
