Skip to main content
Nov 06 2019

Data mining for operational gold

As mines implement more and more digital technologies across their operations, they have to be careful not to miss something more valuable than the ore in the ground — the data being generated by all the digital equipment. Often hidden under the overburden of endless caches of logged processes are nuggets and veins of information that can be refined to operational gold.

OK, OK. I’ll stop with the bad data mining puns — I just couldn’t resist.

But I am seeing a big opportunity for mines in all the data they’re producing. The digitalization of today’s mining equipment and, especially, the adoption of Internet of Things (IoT) sensors and devices is creating isolated lakes of data that can be combined, filtered and analyzed for operational and actionable insights.

This opportunity is often being lost because digitalization in mines usually happens as a by-product of implementing a point solution. Whether it’s autonomous trucks, equipment sensors, personal protection equipment, or environmental monitoring, separate teams don’t normally think of their data in terms of how it can be combined with the data from other teams to generate knowledge, support better decision making and eventually enable the mine’s automation.

The other issue is that there is just so much of it. For a human to sift through all that data looking for stray correlations is simply not worth the effort. But this is where the processing speed and tirelessness of computing platforms comes to the forefront. Using cognitive analytics, machine learning and artificial intelligence, state-of-the-art IT systems are able to filter massive amounts of data and find useful insights such as estimates of how long a piece of equipment is expected to run or under which conditions it is more likely to fail.

This is just one example, but it’s a good one to showcase. Maintenance is one of the key areas where IoT and analytics are combining to change the game. Mining is, of course, hugely asset intensive. The breakdown of key pieces of equipment can be very disruptive. But by analyzing the historical data for a piece of equipment, it is possible for machine learning to create a predictive model for mean time to failure, which it can then modify as real-time data from sensors and maintenance stream in.

This is called ‘predictive’ maintenance. While algorithms can’t actually predict failure, they can narrow the window. This helps maintenance teams to better prioritize resources, and managers can more accurately balance risk against capital spending priorities. If given the right information and parameters, the analytics applications can even use complex risk models to prioritize maintenance vs. replacement — and, they get more accurate over time.

Environmental monitoring is another interesting area for analytics. This is one of the biggest areas of IoT spending by mines. They are putting sensors everywhere – in shafts, in slopes, in ponds – to make mine exploitation more sustainable. And no wonder, mines are risky places and there are any number of different threats to the operation, not to mention the safety of miners.

Take, for example, water. It floods mine shafts and washes away roads, railways and tailing pond dams and dykes. Analytics applications combine hydrological models for the terrain, weather predictions from organizations and live data from environmental sensors to create predictive models about when and where the water is going to flow. Again, this helps to highlight where maintenance should focus and where the mine needs to put its resources.

Finally, there’s the big picture or ‘situational awareness’. Combining all the data coming from the various parts of the mine, including video and audio, with historical data and data from other applications, it’s possible to get a very sophisticated and precise understanding of what is happening almost anywhere, at any time. Of course, the time when this is really critical is when emergencies occur and lives are on the line. But, it also helps for post-incident forensic analysis. Ideally, analytics programs can get smart enough over time to raise flags and alarms before incidents occur.

As I punned badly at the beginning, data mining and analytics processing can uncover operational gold. All kinds of industries are embracing digitalization for just this reason. It’s only in the mining industry, however, where this isn’t just a metaphor.

If you want to learn more about some of the use cases mentioned in this blog, then take a look at our new white paper, “Creating value from IoT and analytics in mining”.  You can download it here. For more information about Nokia’s value proposition for the mining industry, please visit our web page. You can find an overview of our IoT and analytics solutions here.

Share your thoughts on this topic by joining the Twitter discussion with @nokiaindustries using #GoAllwhere #IoT #AI #mining 

About Steve Zakar

As Senior Director for Business Solutions and Strategy, Steve provides vision for all new and existing IoT Analytic Products. For the last six years, he has been with SpaceTime Insight, which was acquired by Nokia in 2018. In his free time Steve likes listening to music, skiing, hiking, racing, fishing, boating, fast cars and bikes, and slow woodworking.

Tweet us at @nokiaindustries