Reinforcement learning–enhanced population-based multi-objective ore blending with iot-observed grade uncertainty in open-pit mines
Authors:
- Azamat UmirzokovAUAzamat Umirzokov
- Shokhjakhon AbdufattokhovSAShokhjakhon Abdufattokhov
- Ryumduk OhRORyumduk Oh
- Shukhratulla OchilovSOShukhratulla Ochilov
- Aidar KuttybayevAKAidar Kuttybayev
- Makhfuza TukhtayevaMTMakhfuza Tukhtayeva
- Kazi BizhanovKBKazi Bizhanov
- Arystan KozhantovAKArystan Kozhantov
- Suhbat NorinovSNSuhbat Norinov
- Samenov GalymzhanSGSamenov Galymzhan
- Ainash KainazarovaAKAinash Kainazarova
Open Access
- 25 August 2026
- journal article
- research article
- Published by Springer Nature in Journal of Engineering and Applied Science
- Vol. 73 (1) , 333
Abstract
Open-pit ore blending must simultaneously control grade-target compliance, economic value, and operating cost with noisy, partial-grade observations from the Mining 4.0 sensing infrastructure. Building on recent population-based multi-objective reinforcement learning (MORL) architectures for ore blending, this work presents an in-depth empirical study of sequential decision-making under IoT-observed grade uncertainty, involving three competing objectives: grade deviation, net present value (NPV), and cost. To overcome preference collapse and insufficient Pareto coverage in single-policy MORL, we systematically extend the specialist-bank approach by introducing uncertainty-weighted multi-objective gradients, a curriculum noise annealing protocol tied to sensor characteristics, and plant-level operational constraints including crusher throughput and haulage costs. We evaluate this extended architecture against single-policy MORL, rolling-horizon evolutionary control, and static evolutionary baselines on open benchmark mine instances under multiple uncertainty levels. The population MORL achieves a higher peak NPV than static NSGA-II on the primary mine instance, matches the performance of rolling-horizon evolutionary control at three orders of magnitude lower inference latency, and generalizes to an external mine instance (Newman1) without retraining, demonstrating practical transfer to a different orebody distribution. These results suggest that population-based MORL is a viable path to realizing real-time, preference-aware blending control in the face of sensor uncertainty and highlight the importance of empirical, constraint-aware analysis when deploying these architectures alongside offline evolutionary optimization.Keywords
This publication has 33 references indexed in Scilit:
- Experimental study and modeling adsorption behavior of a robust cross-linker on carbonate rocks at different temperaturesScientific Reports, 2025
- Unmanned mining fleet Management: A Multi-Objective framework integrating deep reinforcement learning and Internet of ThingsExpert Systems with Applications, 2025
- Deep Reinforcement Learning based real-time open-pit mining truck dispatching systemComputers & Operations Research, 2024
- Methods for calculating the parameters of drilling and blasting operations based on the primary determination of the zones of destruction of the rock massE3S Web of Conferences, 2024
- Study of the Production and Efficiency of Variable and Loading Equipment in the Mining of MineralsE3S Web of Conferences, 2024
- Investigation of the Process of Destruction of a Roain Massif by Borehole Charges with an Axial Air Cavity by the Method of Physical SimulationE3S Web of Conferences, 2024
- Bentonite clays of the Central Kyzylkum and their material composition and engineering and geological properties, UzbekistanE3S Web of Conferences, 2023
- A systematic review of artificial intelligence and data-driven approaches in strategic open-pit mine planningResources Policy, 2022
- A multi-objective deep reinforcement learning frameworkEngineering Applications of Artificial Intelligence, 2020
- A 3D approximate hybrid algorithm for stope boundary optimizationComputers & Operations Research, 2020