Document Type : Original Research Paper

Authors

• Heydar Bagloo ¹
• Mohsen Soleiman Dehkordi ²

¹ Research and Development Unit, Dispatching center, Chadormalu Mining Complex, Yazd, Iran,

² Management Department of Chadormalu Mining Complex, Yazd, Iran

Abstract

Loading and haulage operations in open-pit mining represent a significant portion of overall costs. Among various load and transport systems, the shovel-truck method is favored for its flexibility. Consequently, extensive research has been conducted to optimize this system, resulting in numerous productivity-enhancing methods. However, evaluating the effectiveness of these optimization techniques, particularly in short-term mining activities under varying operational conditions, remains essential. Additionally, understanding how changes in operational conditions impact productivity is important for addressing production fluctuations in daily mining operations. To tackle these challenges, this study uniquely applies advanced machine learning techniques to short-term mining planning, resulting in the development of a real-time Productivity Evaluation Model (PEM) based on supervised learning methods for optimizing truck-shovel operations in open-pit mining. The model, developed and tested using data from a large-scale mining operation in Iran, demonstrated that the Decision Tree was the most effective, achieving an R² value of 0.96. This was closely followed by Random Forest and Gradient Boosting, both with R² values of 0.95. However, the choice of the most suitable learning method may vary depending on the specific dataset and context. The model determines the most appropriate learning method for each dataset within specific mining operations.

Keywords

• Open-pit mining
• Truck
• shovel
• Machine learning
• Supervised learning

Main Subjects

• Loading & Transportation