Prediction of Yttrium, Lanthanum, Cerium and Neodymium Leaching Recovery from Apatite Concentrate Using Artificial Neural Networks

The rare earth elements (REEs) assay and recovery in leaching processes is being determined using expensive analytical methods, ICP-AES, and ICP-MS. This paper presents a neural network model to predict the effects of operational variables on the lanthanum, cerium, yttrium, neodymium recovery in the leaching of apatite concentrate. The effects of leaching times: ten to 40 min, pulp densities: 30 to 50 per cent, acid concentrations: 20 to 60 per cent and agitation rates: 100 to 200 rpm were investigated and optimised on REEs recovery in laboratory at a leaching temperature of 60¦C. The obtained data, on the laboratory optimisation process were used for training and testing of a neural network. A feed-forward artificial neural network with 4-5-5-1 arrangement, was capable of estimating the REEs leaching recovery. Neural network predicted values were in good agreement with the experimental results. The correlations of R = 1 in training stages, and R =  0.971, 0.952, 0.985 and 0.98 in testing stages were results for Ce, Nd, La and Y recovery predictions respectively and that theses values are considerably acceptable. It was shown that the proposed neural network model accurately reproduces all the effects of operation variables, and can be used in the simulation of a REEs leaching plant.