Machine learning solutions

Enterprise mining solutions require us to create and integrate tools that target end-to-end technical processes spanning mining disciplines and stages.

Automated, real-time solutions which leverage machine learning and ‘internet of things’ concepts provide understanding around variances and exceptions as they occur. This then enables better and more timely decisions based on accurate, current operational information which in turn drives continual improvement.

Enterprise systems that incorporate machine learning approaches will drive the need for personnel who are skilled in more than collecting, analysing and reporting data. Interrogating data in the context of a mine’s operational model, and inferring connections between data and processes will be key to unlocking future productivity gains and establishing best practice.

Intelligent systems developed with the knowledge of up and downstream processes can help you better understand and take control of your information. We are all used to the idea that machines can do things that humans can’t - computers can do calculations quicker, robots don’t tire as easily, plotters don’t make silly mistakes and excavators are stronger than a pick and shovel.

The good news is that machines can learn too. And learn fast! With machine learning, the paths that don’t work are discarded and the paths that work are reinforced very quickly. We can reset parameters and a new solution is presented for evaluation. The machine learns what we want and outcomes are improved.

Applications where machine learning can bring immediate benefits include:

  • Enhanced processing and modelling – orebody modelling, grade estimation, grade control
  • Advanced analysis and optimisation – fragmentation, drill and blast planning, scheduling

Maptek expertise

Maptek is now applying machine learning and augmented reality to accelerate tasks such as resource modelling, grade estimation, fragmentation analysis and production tracking. The Maptek Compute Framework enables faster, secure cloud processing and represents a radical improvement over intensive desktop processing and time-consuming data manipulation.

Maptek DomainMCF uses machine learning to generate domain boundaries directly from drillhole sample data for rapid creation of resource models.

Geologists feed in drilling data and obtain domain or grade models in dramatically less time than traditional resource modelling methods.

DomainMCF means projects can be modelled as often as you want. Results are available in minutes and comparable to classical techniques. Geologists remain in control of the process without the onerous preparation work. New data can be added and models regenerated quickly to reflect the current data.

Read more about DomainMCF

Industry partners

Machine learning outcomes are significantly improved when driven in collaboration with industry and technology partners. Maptek is committed to identifying technology partners to develop complementary technologies that enhance the value proposition for our customers. In this way we can share the vision for the future of mining and work more effectively to achieve better outcomes.

Integration between Maptek solutions and PETRA MAXTA targets operational improvement in areas such as ore recovery, blast optimisation and scheduling. The result is that planning is based on a better understanding of the real performance of downstream processes. MAXTA models of the mine performance can add value to the orebody knowledge base, enabling a mine to surpass targets every day.

Case Study: Digital twin models unlock mine value chain optimisation to improve performance