GEMNet - Using Neural Networks to Approximate the Location-Grade Relationship in Mineral Deposits

GEMNet, (Grade Estimation using Mapping Networks), is a grade/reserve estimation software system that uses the artificial intelligence technique known as neural networks to perform reserve estimates from both two- and three-dimensional data samples from a mineral deposit. The system is the result of research carried out over the past three years into the feasibility of using neural networks for reserve estimation at the AIMS Research Unit in the Department of Mineral Resources Engineering at the University of Nottingham. This paper describes the architecture of the GEMNet system including details of the neural network components of the system. The performance of the GEMNet system is then compared to several other widely used reserve estimation techniques on two reserve estimation examples. The results produced by the GEMNet system compare favourably with more conventional estimation techniques, but require fewer assumptions to be made about the form of the data used, and do not require the use of complex mathematical modelling.