Abstract
The application of advanced technology & AI to ore body discovery is one of the ways the mining industry is managing to address the projected supply shortage of key minerals in the coming decade. Massive quantities of siloed geological and geophysical data are used for ore development and exploration. Geoscientists spend up to 80 percent of their time in the search-clean-prepare data cycle, time which could be better spent applying their expertise to model, interpret and prioritize ore body targets. Additionally, the unwieldy volume and variety of exploration data propagates uncertainty across the exploration workflow, which is further hampered by interpretation bias. These issues result in an expensive, time-consuming process for finding new economic-grade mineral resources.
IBM’s Exploration with Watson (ExW) has been streamlining the ore body discovery process for multi-national mining clients since 2017. ExW consists of 1) a centralized digital platform optimized for spatial data and 2) a predictive modelling framework for improved mineral targeting. We discuss how these components have been incorporated into our clients’ existing workflows, vastly reducing the time and uncertainty associated with mineral discovery.
Platform:
The ExW cloud-based platform ingests a multitude of geological and geophysical data in native formats. It establishes spatial relationships among elements and enables powerful first-of-a-kind exploration activities. The platform has proven to perform up to 35x faster than traditional methods in running complex geospatial queries, allowing geoscientists to spend less time in the search-prepare-clean data cycle, and more time modelling, interpreting and prioritizing targets.
Predictive Models:
The ExW predictive model framework is a data-driven approach to mineral targeting that leverages supervised computer vision techniques in deep learning. The formulation is augmented with expert knowledge from exploration teams, enabling the creation of powerful predictive models for virtually any mineral type or geo-characteristic, that are continually retrained and updated with the latest acquired data.