Anything you can do, I can do better…
I’m normally the first to advocate for the presence of people in mining but, in the case of data analysis, ‘machines’ – and by that I mean computers and algorithms – win every time.
Why? Because, simply put, they are better at quantitative tasks than us. Humans love problem solving but can quickly become tired of repetitive tasks. We’re relatively slow at them and prone to making mistakes or missing key information.
To sort through terabytes of would take a human years when a machine could do it faultlessly in minutes.
On the flip side, humans are really good at qualitative tasks. Machines can’t gauge community sentiment and the nuances of a relationship as effectively as we can, for instance.
Because of these differences, it’s highly unlikely that machines will ever replace people entirely in mining… Not in our lifetimes, anyway.
However, exploiting our differences and leaning on technology for certain tasks will not only generate greater efficiencies throughout the value chain and bridge skills gaps in specific jurisdictions, it will also free up the people we do have to work on more complex projects and problems.
In short, it will make our jobs more interesting.
The Intelligent Miner has previously covered machine learning for mineral exploration, but let’s look at some applications in operational optimisation too…
- Newtrax, July 2021: Artificial Intelligence is the Next Frontier of Improving Underground Mining Operations. This article explores how machine-learning algorithms can pull out patterns in data collected around a mine and its fleet to predict problems before they occur, vastly improving the mine’s efficiency, with Agnico Eagle’s Goldex mine in Quebec, Canada as an example.
- Hitachi, November 2019: Integrating machine learning, optimization and simulation to increase equipment utilisation. This one is quite technical – a use case study on open-pit mines, which approaches increasing utilisation by reducing the time spent in non-productive activities using a combination of machine learning with optimisation and simulation. Experiments indicated that the authors’ approach could improve overall equipment effectiveness (OEE) by 10% when compared to the current dispatching software.
- JKMRC, November 2019: Artificial intelligence and machine learning in mineral processing – challenges and opportunities. A video of a seminar by researchers at the Julius Kruttschnitt Mineral Research Centre (JKMRC) at the Sustainable Minerals Institute (SMI) at the University of Queensland in Australia. Some of the topics covered in this in-depth seminar include the capabilities and limitations of AI and machine learning in mineral processing, with examples such as using it to detect a ball mill overloading.
- ESI, June 2021: Predictive Maintenance of Conveyor Belts with Digital Twins. An interview with ESI’s Jörg Arloth that looks at how data-based methods such as big data and machine learning can enable smart services for applications like predictive maintenance, with the example of a digital twin of a belt conveyor.
- PETRA Data Science, May 2019: Can machine learning predict tailings dam failure? In 2019, PETRA managing director Dr Penny Stewart co-hosted an Austmine Machine Learning Workshop in Brisbane with METS Ignited and the Australian Institute for Machine Learning. The workshop participants voted that predicting the failure of tailings dams was the most valuable idea for potential machine-learning applications. The article goes on to ponder whether this will soon be possible, with input from Maptek CEO, Peter Johnson, and TRE Altamira CEO, Alessandro Ferretti.