An application of artificial neural network for prediction of densities and particle size distributions in mineral processing industry

This paper demonstrates an application of artificial neural network (ANN) for determination of underflow and overflow densities of hydrocyclone separators. The discussions are extended and further results are presented for the prediction of particle size distributions in the underflow and overflow streams. The fit of the experimental results against the predicted results are illustrated and a statistical analysis is made. It is shown that, once the history of the operations are known, the ANN proves to he a useful tool for predicting future separation efficiencies. This approach has a potential to eliminate the need for installation of expensive on-line instruments for density measurements and particle size analyses. This approach can be applied in similar situations in the mineral processing industry