Bradshaw Research Institute for Minerals and Mining

In the mining industry, safety is absolutely paramount. We cannot have algorithms that are only correct 90 percent of the time. – Bhushan Gopaluni 

If data is the new gold, Bhushan Gopaluni has figured out how to cash in. Working out of the DAIS Lab at the University of British Columbia, Gopaluni is using artificial intelligence to transform a broad range of sectors – mining, oil, gas, paper, pharmaceutical, etc – that are traditionally dependent on physical labor and machinery. Tapping into the data resources these industries generate, his team employs A.I. to sift through rich datasets to diagnose problems, optimize processes, and ultimately automate physical plants to create safer, more efficient industrial environments.

“These industries possess vast, underutilized troves of data,” says Gopaluni, a researcher at UBC’s Bradshaw Research Institute for Minerals and Mining (BRIMM). “Our work starts where the data exists. Simply put, how do we extract actionable knowledge or information from the existing data? And we’ve discovered that harnessing this information can unlock unprecedented insights, easing economic and competitive pressures.” 

This approach is particularly relevant in the mining and extraction industries, where raw materials undergo a series of physical processes on their way to becoming finished products. Each step generates critical data – temperatures, pressures, flow rates, and other properties – which can then be analyzed for inefficiencies that are hampering the process or the quality of the finished product.

“You can use all of the information you’re generating through the sensors to understand how these processes behave, how they change dynamically.” says Gopaluni. “If you have that understanding, you can use that to control them, to maneuver them in such a way that you are maximizing the efficiency of the process, maintaining quality of your products, minimizing energy consumption, and maximizing your profits.”

One of the lab’s key innovations is the development of “digital twins” – dynamic computer models that serve as virtual replicas of real plant operations. The lab sifts through the historical data to build these virtual models, identify any faults, and diagnose their causes. Using predictive analytics, the team then uses the models for a post-mortem analysis that will inform the design of more optimal processes. 

But ‘big data’ isn’t the same thing as ‘good data’. “The data that we get often has a lot of undesirable features,” says Gopaluni. “It may have outliers, missing data, things that are not measured properly, data from faulty sensors, a bias in the measurement – so we have to look for all of those data and remove them, or input new data points to clean it in such a way that we can use it in a machine learning algorithm.”

“If your goal is to identify objects in images and you make a mistake, nothing happens,” he adds. “But in the mining industry, safety is absolutely paramount. We cannot have algorithms that are only correct 90 percent of the time. We need guaranteed results.”

With the emergence of A.I. as the new innovation frontier, Gopaluni’s holistic approach to data isn’t just pragmatic, it’s inevitable. It’s hard to imagine a future in which industrial processes aren’t just automated, but intelligent – autonomous plants with connected assets, self-diagnosing systems, adaptive maintenance free controllers, and A.I.-enhanced operational intelligence – a significant stride towards higher automation, efficiency, and safety in the mining sector.