Newtrax Technologies says it has applied machine-learning algorithms to help Agnico Eagle Mines’ Goldex mine predict mobile equipment maintenance issues up to two weeks in advance.
With the two companies already having an existing relationship at the mine, in Quebec, Canada, Newtrax was approached in the fall to discuss the data Agnico had collected from sensors over the past six years. This amounted to 10 billion data points, according to Newtrax.
“This data was exactly what was needed to apply machine-learning algorithms in order to predict mobile equipment maintenance issues at least two weeks before they were supposed to happen,” Newtrax said.
Daniel Pinard, Team Lead, Special Projects with Agnico Eagle, said this predictive Newtrax AI solution allowed the company to intervene before incurring serious problems that could potentially break vehicle engines.
“Through the use of machine-learning algorithms with Newtrax, we were recently able to analyse an engine that had a potential problem and we saved it from failing. This helped Goldex mine avoid serious damage on that engine which saved them C$85,000 ($63,610).”
The Newtrax AI solution is unique in three ways, according to Michel Dubois, VP QA & Artificial Intelligence at Newtrax, “first, Newtrax has years of unique data that is extremely well suited for machine learning (ML)”.
This creates a source of training data for ML that is unique in the world, with the data growing every time a mining company decides to join in, he said.
“Second, we have a unique AI team who knows how to generate actionable results using existing AI algorithms. And, third, we have a unique approach where our AI specialists go underground and focus on quick wins, and they leverage those existing algorithms to solve high-value problems.”
This is the first ever applied case study for ML in the underground hard-rock mining industry with a defined return on investment, according to Newtrax.
Newtrax said it worked with artificial intelligence and ML researchers such as IVADO to apply existing algorithms to the data collected in mine sites.
