Prediction of results of floatation process using artificial neural networks

In this paper, the use of artificial neural networks in the flotation process is presented and analyzed. Apart from utilization of flotation in the mineral processing, deinking flotation is used for separation of ink particles from the cellulose fibers. The time of deinking flotation, pH value, reagents such as oleic acid, oleic acid with CaCl2, oleic acid with AlCl2 as well as concentration of reagents were used as input for the network. The recovery of toner particles in froth was used as output data for the network. The neural network demonstrated high correlation coefficients during training (0.97), validation (0.96), and testing (0.94), and subsequent accuracy testing confirmed these results.