Mineralogical\nAnalysis of Solid-Sample Flame Emission\nSpectra by Machine Learning

Solid\npreconcentrated ore samples used in pyrometallurgical\ncopper\nsmelters are analyzed by flame emission spectroscopy using a specialized\nflame optical emission spectroscopy (OES), system. Over 8500 complex\nspectra are categorized using an artificial neural network (ANN) that\nwas optimized to have 10 hidden layers with 40 nodes per layer. The\nANN was able to quantify the elemental content of all samples to within\nbetter than 1.5 mass% and was able to identify the prevalent minerals\nto within better than 2.5 mass%. The flame temperature was obtained\nwith an uncertainty of σ < 3 K and the particle size to within\n2 μm. The results are found to be superior to those obtained\nto a nonlinear partial least-squares fit model, which is equivalent\nto an ANN having no hidden layers.