Analysis and prediction of the thiourea gold leaching process using grey relational analysis and artificial neural networks

Abstract

The thiourea (TU) leaching of gold from refractory ores can be considered an alternative to cyanidation. However, the high reagent consumption causes an increase in cost, which seriously limits its use. In order to effectively reduce the TU consumption, it is necessary to analyze the influencing parameters of gold recovery and TU consumption and apply them to the prediction of the TU leaching process. This paper investigated six potential influencing parameters and used grey relational analysis (GRA) to analyze the relational degree between each parameter and gold recovery and TU consumption. Then, the artificial neural network (ANN) model was established to simultaneously predict the gold recovery and TU consumption in the TU gold leaching process. The results of the GRA indicated that the leaching time, initial pH, temperature, TU dosage, stirring speed, and ferric iron concentration were all well related to the gold recovery and TU consumption. Therefore, the incorporation of these parameters can significantly improve the ANN model validation. The predictive results noted that the prediction accuracy of gold recovery varied from 94.46% to 98.06%, and the TU consumption varied from 95.15% to 99.20%. Thus, the predicted values corresponded closely to the experimental results, which suggested that the ANN model can accurately reflect the relationship between the operational conditions and the gold recovery and TU consumption. This prediction method can be used as an auxiliary decision-making tool in the TU gold leaching process, and it has broad engineering application prospects in engineering.