SPATIAL ESTIMATION OF A COAL RESOURCE USING ARTIFICIAL NEURAL NETWORKS

Spatial estimation can be described as the estimation of the variable in unknown locations under study respect to the known locations using available data. In mining, spatial estimation of unknown target location plays a vital rolesince the data collection is challenging and only limited number of data is available. Kriging is most widely used spatial estimation industry standard method. In order perform the estimation using kriging; good command on variogram modelling is required. Also, stationarity assumption in kriging has to be valid which is not common case. In estimations, a method which does not use variogram and does not have strong assumptions like stationarity providesadvantage over the existing estimation methods. This study aims to use Artificial Neural Networks (ANN) for spatial estimation purpose. ANN used in spatial estimation of the Lower Calorific Value (LCV) of a coal resource. Neuralnetwork of fifteen hidden neurons are trained with 226 LCV data using training algorithm. The locations where no data is available is estimated using the model constructed previously. Matlab ® Neural Network Toolbox is used in allmodel construction and spatial estimation steps. Results of the estimation is compared with the original LCV data which shows that ANN can be used in LCV estimation.