Modeling of Phenol Extraction from Wastewater Using Intelligent Techniques

In this study, the extraction of phenol from wastewater was simulated using intelligent methods which include multi-layer perceptron, radial basis functions network, and support vector regression. To design the network structure and to train and test it, 184 experimental data sets were used. Inputs to the network consisted of organic–aqueous volume ratio, rotor speed, temperature, pH, and time while extraction efficiency was the output. Root mean square error and correlation coefficient were used in all the three models as network performance and network stop criteria. Comparison of the results obtained from the three models revealed that the support vector regression was the best model with a correlation coefficient of 0.684 and a root mean square error of 0.99. Moreover, model results showed good agreement with experimental data. Optimal process operational parameters included an organic to aqueous volume ratio of 0.22, a rotor speed of 350 rpm, a temperature of 22.86 °C, a pH equal to 7.5, and an agitation time of 15.86 minutes; the corresponding extraction efficiency was obtained to be 96.35.