Neural network prediction of bioleaching of metals from waste computer printed circuit boards using Levenberg‐Marquardt algorithm

Abstract

AbstractThe applicability of artificial neural network (ANN) to predict the bioleaching of metals using from computer printed circuit boards (CPCB) and the influence of process parameters were studied. The influence of process parameters initial pH (1.6‐2.4), pulp density (2%‐13%), and the initial volume of Inoculum (5%‐25%) were investigated on the rate of bioleaching of metals from CPCB. Network inputs were fed as initial pH, pulp density, and inoculum volume and with the extraction of Cu, Ag, and Au as output. The ANN was developed using the Levenberg‐Marquardt algorithm and trained for modeling and prediction. The most fitting architectures for Cu, Ag, and Au were [4‐5‐5‐2‐1], [4‐7‐5‐2‐1], [4‐7‐1‐1‐1] trained with Levenberg‐Marquardt algorithm, respectively. The R values were observed to be 0.996, 0.997, and 0.993 for Cu, Ag, and Au extraction predictions, respectively. The genetic algorithm model defined by ANN was used to achieve maximum extraction rates for Cu, Au, and Ag. The predicted data showed that there is a great capability of using ANN for the prediction of Cu, Ag, and Au extraction from CPCB through bioleaching process. Hence, the ANN model can be used to control the operational conditions for improved metals extraction through bioleaching.