Optimisation and Prediction of Co3+-Enhanced Dissolution in Sulfuric Acid

The paper explains the main effects of input leaching parameters, such as addition of a reducing agent, acidity, agitation, temperature, and leaching system potential, in relation to observed and predicted yields of sought-after cobalt. An artificial neural network coupled with graphical probability networks was developed wherein the abovementioned input parameters were used to predict the percentage of dissolved cobalt. Furthermore, similarities and differences between the Bayesian and neural network approaches in terms of their key outcomes are compared with respect to study of the dissolution of cobalt in sulfuric acid. Key results indicate that there is a need to use neural network results to predict conditions for dissolution of cobalt, while using Bayesian network to explain inferences about conditions of the input parameters, thereby achieving realistic predictions.