An AI-Powered System for Predicting Energy Consumption in Ore Comminution Operations

Ore comminution, a highly energy-intensive stage in mining, necessitates optimization for economic and environmental benefits. Traditional energy prediction methods, reliant on empirical and physical models, lack precision in modeling particle size distribution–energy relationships. This study proposes an AI-driven system integrating real-time particle size recognition and energy consumption prediction to enhance comminution efficiency. The YOLOv8 algorithm enables real-time ore size detection, combined with BoT-SORT multi-object tracking to quantify particle size distribution. A synchronized data acquisition system captures ore video and power meter readings, providing high-quality datasets. During processing, optical character recognition (OCR) extracts meter data, while timestamp alignment ensures synchronization between particle size and energy metrics. The proposed method innovatively integrate KAN and DeepSet to establish quantitative relationships between particle size distribution and crusher energy use. Experimental results demonstrate that the proposed model significantly improves energy prediction accuracy and exhibits strong generalizability. By enabling real-time monitoring and predictive energy optimization, this system offers an intelligent solution for green and smart mining practices. The innovation lies in merging AI algorithms with real-time data acquisition, advancing energy-efficient comminution and providing a novel technical pathway for sustainable mining. This research underscores the potential of AI-driven systems to transform traditional industrial processes, balancing operational efficiency with environmental stewardship.