Chapter (24): Application of Machine learning for geometallurgical modelling

CHAPTER DESCRIPTION

The application of Machine learning in mining industry has risen significantly during the last decade. Automated processes made it possible to explore a larger dataset and solve problems faster, find hidden relations between variables easier, extract meaningful information from an unstructured dataset and reduce human error. Geometallurgy on the other hand as an interdisciplinary approach combines geology, mining and metallurgy to maximize Net Present Value and optimize mine production planning. With all the challenges miners are facing today, implementing such an
approach could help mitigate risks concerning productive orebody management. Building a spatial model that predicts the metallurgical parameters, improves the capability to recognize the bottlenecks in mine production and highly influences the economic value of the project. Although many fields take advantage of machine learning and deep learning to develop predictive models, so far most of the 3d geometallurgical modelling has been constructed with classic statistical modelling methods and automated approaches have been underutilized for this purpose. In this chapter, we first discuss the possibilities of implementing such an approach, its ability to solve problems, and its probable drawbacks. Then for the next step, we will explore some recent researches conducted in this field.

CHAPTER CONTENT

• Geometallurgy, an essential element in the mine value chain
• representative sampling
• Geometallurgical domaining
• data types that are used for geometallurgical modelling
• Metallurgical test and Analysis
• Data pre-processing
• Machine learning and deep learning approaches for gold geometallurgical modelling (explaining methods such as SVM,
• decision trees, random forest, ANN)