Advances in Research

Mining operations increasingly require intelligent control strategies capable of balancing productivity with energy efficiency, water stewardship, and operational stability. While Mining 4.0 technologies have enhanced automation and monitoring capabilities, most existing systems remain largely reactive and lack intrinsic self-awareness and sustainability-driven decision logic.

This study proposes an Eco-Cognitive Mining System (ECMS), formulated as a hybrid digital twin–surrogate-based, multi-objective cognitive control framework for mineral processing systems. The framework integrates digital twin–based state perception, surrogate eco-physical modeling, and reinforcement learning to enable adaptive, sustainability-aware decision-making under dynamic operating conditions. Energy consumption, water usage, operational stability, and metallurgical recovery are embedded directly within the cognitive optimization process.

Simulation-based evaluation on a grinding–classification circuit demonstrates that ECMS reduces disturbance recovery time by approximately 45% and lowers peak system deviation by about 14% compared to conventional control. Cognitive learning converges within 50–60 episodes, representing a 45–50% improvement in learning efficiency over conventional reinforcement learning. Sustainability trade-off analysis indicates reductions of approximately 11% in energy intensity and 16% in water consumption while maintaining metallurgical recovery. System self-awareness, quantified using a Self-Awareness Index, increases from approximately 0.72 to 0.90, with statistically significant improvement (p < 0.001). These results confirm that ECMS enables self-aware, adaptive, and sustainability-driven mineral processing, providing a practical pathway toward Mining 5.0 systems.