The use of machine sensors in mining and mineral processing operations is often presented alongside computing and robotics as advances that can improve on human capabilities. Perhaps before trying to replace the person, it is necessary to first examine and replicate what made them special.
Professor Dr Elisabeth Clausen, head of Advanced Mining Technologies at RWTH Aachen University, told Mining Magazine about collaboration work she has done on the development of sensor technologies for material characterisation on conveyor belts.
Can you tell us more about this project?
The specific idea in this project was to identify and characterise materials directly on the conveyor belt using structure born sound in a high frequency spectrum, particularly identifying and sorting gypsum and anhydrite during the excavation and transportation stage.
This is an important topic for Germany as almost half of our gypsum comes as a by-product from coal plants and will need to be replaced by gypsum from primary resources which may contain unwanted anhydrite.
The wider goal is also to start a deeper integration of the mining and mineral processing stages, while getting personnel out of danger areas and becoming more energy and resource efficient.
By tackling this at the excavation and transportation stage we can generate information for the subsequent processing stage as well as incorporate feedback into the mining process. We can gain a better understanding of the deposit, adjust our extraction and being more selective to create less waste materials and increase the recovery.
What else attracted your team to work on machine sensor projects?
The idea is to take human experience, and link this to the sensor technology and to the machinery. We want to make machinery that's able to feel, to hear and to see similar to a human.
The idea for this project is based on the fact that experienced operators in the field have a sensory understanding of materials, they can hear the changes in the characteristics of material that is being handled and processed.
First we need to try and recreate these abilities on a physical level, or rather to ‘hear' with the sensor technology. Therefore, we are using high frequency methods of structure-born sound.
This highlights an important factor not just in sensor technology but also in the use of data visualisation. In any advances, a close link must be kept with the experienced people in that field. This "domain knowledge" needs to be integrated into data processing.
And can these types of technologies increase safety levels by removing human operators from riskier parts of the mining chain?
Certainly, because for example we are also working on technologies that can be used for material- and deposit-boundary layer detection at the face, or other sensor technologies for inspection purposes.
This is particularly true for mining operations where you need to follow the deposit. Often you need the experience from the people who ‘feel' or "see" where the deposit is headed. If we had the technology that was able to see these boundary layers, we could bring this expertise into the machinery and create more remotecontrol stations.
Often we face the situation, for example with a wheel-loader, that very good operators can work by ‘feeling' about the base, the material and the direction of travel. The more ways we find to incorporate these perceptions into machinery and sensor technology, the better. The more these processes are then automated, the more people can be removed from dangerous operations.
Digitalisation is not something that would only benefit larger, more advanced mines. In smaller operations we cannot expect full automation, but digital solutions can certainly be used to great effect for example in training and safety procedures.
Does the sheer amount of data being generated by today's solutions make reducing complexity a major challenge?
A lot can be learned from data but it is of most value if combined with people who know the fundamentals of mechanical systems, the process and the domain - data must really bring an added value to decision makers.
With more data we can enrich our systems, but we need to include the experts from both data science and from the mine side. The solution cannot be just "increasing the haystack to find the needle".
Perhaps the operator faces the greatest challenge as they focus almost simultaneously on the information coming from for example the mining, the maintenance and the geology. They are real "pragmatic information aggregators". It will be a real challenge to integrate all these perspectives in a system on our way to autonomous mining.