Meta-Model Development for Mine-to-Mill Optimisation Using AI and Simulation

Date

2025

Authors

Nobahar Ghezeljehmeidan, P.
Xu, C.
Dowd, P.

Editors

Advisors

Journal Title

Journal ISSN

Volume Title

Type:

Conference paper

Citation

Proceedings of the 11th World Congress on Mechanical, Chemical, and Material Engineering (MCM'25), 2025, pp.MMME 133-1-MMME 133-10

Statement of Responsibility

Pouya Nobahar, Chaoshui Xu, Peter Dowd

Conference Name

11th World Congress on Mechanical, Chemical, and Material Engineering (MCM) (19 Aug 2025 - 21 Aug 2025 : Paris, France)

Abstract

The current demand for mineral resources is higher than it has ever been, and it is expected that, over time, the quality of future resources will decline, and they will become more difficult to extract. Routinely collected on-site data from various mining stages are often neglected in mining operations and are not being used to improve the value of the mining chain. To address this issue, mining companies need to increase the efficiency of their mining processes to achieve sustainable production by using innovative solutions. The primary purpose of the study presented here is to develop an integrated knowledge-based system using advanced AI techniques to simulate, monitor, assess, and optimise mining processes from blasting to downstream products. In this study, publicly available data from the Barrick Cortez Mine in Nevada, USA, was used to model the entire mining process from blasting to SAG mill by using Orica’s Integrated Extraction Simulator (IES) platform. The comparison of real data from the mining site with simulated data on the IES platform demonstrates that the modelled operations closely match the real data. Thirteen parameters related to blasting, screens, crusher, and SAG mill were considered. Given the computational infeasibility of testing all combinations, three million scenarios were simulated to identify key performance drivers. Machine learning models—including linear regression, decision trees, random forest, and XGBoost—were evaluated to determine the most effective for predicting outcomes. The next step involved using input scenarios and outcomes to investigate key features and interpret results using feature importance and SHapley Additive exPlanations (SHAP) techniques, respectively, as powerful tools for determining the influence of individual features of the models. The findings highlight the potential of AI-driven meta-models to enhance decision-making, reduce operational costs, and improve resource usage in mining operations.

School/Discipline

Dissertation Note

Provenance

Description

Session: Mineral Processing. Paper No. MMME 133. There are 4 conferences included in the MCM Congress: HTFF 2025, ICMIE 2025, MMME 2025 and ICCPE 2025.

Access Status

Rights

© 2025, International Aset Inc. - All Rights Reserved.

License

Call number

Persistent link to this record