<b>Prediction of Optimized Chalcopyrite and Cobalt III Oxide Microwave Pre-processed System Solubilization Behavior in Acid Medium by Artificial Neural Network</b>

Five different architectures of Multilayer Perceptron (MLP) ANNs – MLP 4-4-2, MLP 4-6-2, MLP 4-9-2 (two variants), and MLP 4-10-2 – were developed and compared. Input parameters included microwave power (280-700 W), treatment time (2-8 min), particle size (45-106 µm), and stoichiometric ratio (0-3), while copper and cobalt solubilization yields served as output variables during leaching of pre-processed ores at pH 1.5 in 2 hours.</p><p dir="ltr">Experimental results showed solubilization yields ranging from 4.16% to 58.92% for copper and 0% to 96.03% for cobalt.

In optimal conditions, solubilization yield of 62.56% for copper was obtained after leaching of pre-processed sample at 700W, 4 minutes, 53 microns and heterogenite/chalcopyrite ratio of 1:1; while the yield of 85.94% for cobalt was obtained at 700 W, 6 minutes, 45 microns and 1:1 ratio. Sensitivity analysis revealed that the stoichiometric ratio was the most influential parameter, with an average relative importance of 11.52, followed by particle size (1.90), microwave power (1.93), and treatment time (1.33).