Unplanned dilution and ore loss prediction in longhole stoping mines via multiple regression and artificial neural network analyses
H Jang, E Topal, Y Kawamura
Unplanned dilution and ore loss directly influence not only the productivity
of underground stopes, but also the profitability of the entire mining
process. Stope dilution is a result of complex interactions between a
number of factors, and cannot be predicted prior to mining. In this study,
unplanned dilution and ore loss prediction models were established using
multiple linear and nonlinear regression analysis (MLRA and MNRA), as
well as an artificial neural network (ANN) method based on 1067 datasets
with ten causative factors from three underground longhole stoping
mines in Western Australia. Models were established for individual mines,
as well as a general model that includes all of the mine data-sets. The
correlation coefficient (R) was used to evaluate the methods, and the
values for MLRA, MNRA, and ANN compared with the general model were
0.419, 0.438, and 0.719, respectively. Considering that the current
unplanned dilution and ore loss prediction for the mines investigated
yielded an R of 0.088, the ANN model results are noteworthy. The
proposed ANN model can be used directly as a practical tool to predict
unplanned dilution and ore loss in mines, which will not only enhance
productivity, but will also be beneficial for stope planning and design.
Keywords: stoping, unplanned dilution, ore loss, artificial neural network.