Artificial neural network approach for modeling of conversion rate of refractory gold concentrate oxidation by nitric acid

Considering that the pretreament of refractory gold concentrate is a complex and nonlinear process, it is necessary to develop a new route for optimum control and management. In this study, artificial neural network (ANN) was adopted. Particle size (50-335 ¿m), reaction temperature (25-85°C), nitric acid concentration (10-30%, wt.), stirring speed (400-800 rpm) and reaction time (10-90 min) were chose as input variables, while conversion rate was chose as output target. The tansig function was used as the transfer function in the only hidden layer with 11 neurons and logsig transfer function at output layer. A feed forward neural network model with back propagation algorithm was developed to predict the conversion rate of refractory gold concentrate based on 125 experimental sets obtained in a laboratory batch study. The mean squared error (MSE) become stable at 0.000998771 when the numbers of epochs reach 253. The model was evaluated by comparing the simulated results with the experimental values and was found to be in good agreement with a correlation coefficient of 0.99.