Application of artificial neural network (ANN) for prediction of acenaphthene (ACN) removal in wastewater

In this study, artificial neural network (ANN) was used to predict the adsorptive removal of acenaphthene (ACN) onto activated carbon (AC). ACN is one of the hazardous polycyclic aromatic hydrocarbons (PAHs) that need to be removed from wastewater on priority basis. Prediction and optimization of the ACN adsorption process is most needed to reduce the wastewater treatment process cost and material. For the experiments, the pH, AC dosage, and contact time were taken as the input datasets for ANN training. As a training algorithm, the Levenberg Marquardt backpropagation (LMBP) algorithm was adopted. The training network was run 20 times to get a distinct output and to achieve a minimum mean square error (MSE) and correlation R-value near 1. The R-value near 1 shows a good correlation between the targeted outputs and model predicted outputs. The results showed that the model can accurately estimate ACN removal under various operating parameters as the R values for the training, validation, and testing were 0.97637, 0.99999, and 1, respectively.