Author(s)

Samaneh Liaghat, Nadhir Al Ansari

Environmental and Natural Resources Engineering Department, Luleå University of Technology, Luleå, Sweden.

The nature of rock fragmentation affects the downstream mining processes like loading, hauling, and crushing the blasted rock. Therefore, it is important to evaluate rock fragmentation after blasting for choosing or designing optimal strategies for these processes. However, current techniques of rock fragmentation analysis such as sieving, image-based analysis, empirical methods or artificial intelligence-based methods entail different practical challenges, for example, excessive processing time, higher costs, applicability issues in underground environments, user-biasness, accuracy issues, etc. A classification model has been developed by utilizing image analysis techniques to overcome these challenges. The model was tested on about 7500 videos of load-haul-dump (LHD) buckets with blasted material from Malmberget iron ore mine in Sweden. A Kernel-based support vector machine (SVM) method was utilized to extract frames comprising loaded LHD buckets. Then, the blasted rock in the buckets was classified into five distinct categories using the bagging k-nearest neighbor (KNN) technique. The results showed 99.8% and 89.8% accuracy for kernel-based SVM and bagging KNN classifiers, respectively. The developed framework is efficient in terms of the operation time, cost and practicability for different mines and variate amounts of rock masses.

Rock Blasting, Load Dump Haul, Image Analysis, Artificial Intelligence, Mine Productivity

Liaghat, S. and Ansari, N. (2025) Erratum to “Rock Fragmentation Classification Applying Machine Learning Approaches” [Engineering, 15 (2023) 378-395]^*. Engineering, 17, 1-18. doi: 10.4236/eng.2025.171001.