Technical Session - B4 (TS-B4) | Understanding the Rock Mass – Development of Deposit-Specific Machine Learning Techniques for Geotechnical Domaining
Abstract
The time it takes to bring a deposit from initial discovery to operation is often a decade, or longer, with multiple stakeholders
tasked with collecting drillhole data. During this time, much effort is spent on data collection to improve the understanding
of the resource, with limited focus on geotechnical data collection until the project is at a more advanced stage. The
automated classification of core photographs using machine learning (ML) and computer vision techniques is, however ,
becoming more common to overcome geotechnical data gaps. However, a one-size-fits-all classification is often ineffective
at appropriately characterizing the geotechnical conditions.
This paper outlines SRK Consulting’s established ML workflows that are actively being applied to a variety of deposit styles
globally. This approach to automated characterization of core images provides the potential to leverage already existing
data at earlier stages in a mining project. With proper consideration to deposit context at the outset of the classification
development, this may produce more confident structural or geotechnical domain models and highlight areas of
geotechnical concern sooner than would be possible using conventional geotechnical characterization methods.
Authors
- Andrew LeRiche | Senior Rock Mechanics Engineer| SRK Vancouver
- David Bonneau | Consultant (Rock Mechanics) | SRK Vancouver
- Samson Tims | Senior Geologist | SRK Vancouver
- Ed Saunders | Principal Rock Mechanics Engineer | SRK Vancouver
