The CSIRO Mine Informatics project is working with the Mining Industry to address productivity improvements by the application of intelligent informatics methods. The main activities are: mapping and characterising major voids in underground mines using remotely operated flying vehicles, and virtual modelling of mining and mineral processing operations, enabling automation of the discovery of opportunities for improving efficiency through the application of machine learning techniques.
Mine Virtualisation and Optimisation
Mine virtualisation involves the creation of different forms of digital models of mine design, structure, operations, mine and mill process flow charts, value chains, etc.. Optimisation then uses the multidimensional, holistic system representation as input to the automated discovery of opportunities for improved efficiency.
The optimisation process uses digital models together with data from a combination of sensors and enterprise software as inputs to multi-paradigm machine learning techniques that are selected, applied and evaluated (via reference to real data and simulation studies) by an automated meta-learning process.
The project is also developing interfaces and visualisations to make intelligent optimisation comprehensible to a wide range of users within mines, such as mine managers and supervisors, planners, maintenance controllers and procurers.
Remotely Operated Void Mapping
Mines contain spaces that cannot be accurately mapped by conventional methods because they are too unstable for human entry, or are old and have unknown stability.
Mapping is nevertheless of potentially great value, for assessing the effectiveness of blasting and stoping, for understanding the mineralogy and grades of older mine spaces, and for understanding potential and emerging safety issues in mine voids that cannot be entered, such as water accumulation, or progressive collapse of higher workings.
Automated flying vehicles can enter these spaces, flying above rocky or flooded floors, and flying up to survey high roof areas and adjoining passages.
Mapping can be undertaken by 3D cameras and laser scanners.
We are also developing instrumentation systems for characterising the mineralogy of these spaces, providing data for planning ongoing stoping, for mine closure planning and for evaluation of the benefits of reopening older mine areas for mining.
Mineralogical characterisation will be accomplished by data fusion and the development of classifiers using machine learning techniques to process a combination of new and existing sensor data.
These techniques can also be used for assessing stability and movement in open cut mines.
For more information, contact: firstname.lastname@example.org