Akademik Veri Yönetim Sistemi
Mining, Metallurgy and Exploration, 2025 (SCI-Expanded)
Rock displacement velocity (RDV) is a critical factor influencing the efficiency of rock fragmentation and displacement during blasting operations. This study explores the relationship between geological joint characteristics, blast design parameters, and RDV in sedimentary rock mines using an innovative UAV-enabled field data collection approach. A comprehensive dataset from 82 production blasts was analyzed, incorporating key variables such as joint angle, joint set number, burden-to-spacing ratio (B/S), stemming length, firing pattern, and explosive quantity. Advanced machine learning techniques, including adaptive boosting (AdaBoost), decision tree, Gaussian process regression, support vector regression, and random forest were employed to predict RDV. Among these, random forest demonstrated superior predictive accuracy, achieving an R2 score of 0.97 in training and 0.60 in testing, along with minimal error. Performance was further validated using metrics such as variance accounted for (VAF), mean absolute error (MAE), root mean square error (RMSE), and correlation coefficient. Additionally, data analysis techniques, including 2D kernel density estimation and Spearman rank correlation, were applied to assess feature interactions. The Spearman correlation revealed a strong positive relationship between joint set number (JN) and rock permeability (RP) (0.73), as well as a moderate correlation between RDV and B/S ratio (0.45). These findings highlight the significant role of rock discontinuities in optimizing explosive energy utilization and fragmentation efficiency. This research underscores the importance of integrating geological joint properties into blast design and demonstrates the effectiveness of ensemble learning models in accurately predicting rock displacement velocities. The insights gained can enhance blast planning, improve fragmentation outcomes, and contribute to safer and more efficient mining operations.