Language selection

Search

Detecting deep mineral deposits with deep learning for resource extraction

Mineral deposits are the source of important materials for trade within the Canadian economy, including critical minerals, which face a low or non reliable supply and high demand. As major discoveries of near-surface mineral deposits are declining globally, new methods are needed to detect economical deposits at great depths. However, this is challenging due to the relatively small size of ore deposits, the limited number of existing geological data at depth, and limitations of the geophysical methods used for their detection. Machine learning can aid in developing better models for the prediction of rock type and economical mineral deposit locations for extraction purposes without engaging in time and resource-intensive approaches.

Project objectives

Machine learning/deep learning algorithms can be applied to help determine the probability of the presence of valuable mineral deposits and rock type in Canadian mining camps. This project will establish deep learning models linking relationships between 3D seismic data and rock types at the Lalor mine site in Snow Lake, Manitoba, and use those models to predict rock types and mineral deposits in areas where only seismic data exist. This pilot project will determine the data types and the optimal network architecture needed to successfully apply this approach to other locations with great economical potential.

Expected results

The development of new artificial intelligence-informed models will help advance the competitiveness and economic advantage of the Canadian mining industry. Many potential machine learning/ deep learning approaches will be developed. These models will allow for unprecedented predictive capabilities in relation to mineral deposits. Furthermore, the project will provide several benefits, including:

  • An open source machine learning workflow
  • Opportunities to scale-up and replicate project outcomes across Canadian mine sites
  • Enhanced domestic and international collaboration opportunities for mineral development

Data sources

  • Available Lalor mine data comprises:
    • An extensive drillhole database with rock types
    • A detailed 3D geological model constructed from the drillhole data
    • 3D seismic data set

Sector


Collaborators and Partners

  • MILA – Québec Artificial Intelligence Institute

Contact

Gilles Bellefleur

Report a problem on this page
Please select all that apply:

Thank you for your help!

You will not receive a reply. For enquiries, contact us.

Date modified: