Estimation of 2,4-dichlorophenol photocatalytic removal using different artificial intelligence approaches

Photocatalytic degradation is one of the effective methods to remove various pollutants from domestic and industrial effluents. Several operational parameters can affect the efficiency of photocatalytic degradation. Performing experimental methods to obtain the percentage degradation (%degradation) of pollutants in different operating conditions is costly and time-consuming. For this reason, the use of computational models is very useful to present the %degradation in various operating conditions. In our previous work, Fe3O4/TiO2 nanocomposite containing different amounts of silver nanoparticles (Fe3O4/TiO2/Ag) were synthesized, characterized by various analytical techniques and applied to degradation of 2,4-dichlorophenol (2,4-DCP). In this work, a series of models, including stochastic gradient boosting (SGB), artificial neural network (ANN), adaptive neuro-fuzzy inference system (ANFIS), the improvement of ANFIS with genetic algorithm (GA-ANFIS), and particle swarm optimization (PSO-ANFIS) were developed to estimate the removal percentage of 2,4-DCP. The model inputs comprised of catalyst dosage, radiation time, initial concentration of 2,4-DCP, and various volumes of AgNO3. Evaluating the developed models showed that all models can predict the occurring phenomena with good compatibility, but the PSO-ANFIS and the SGB models gave a high accuracy with the coefficient of determination (R-2) of 0.99. Moreover, the relative contributions, and the relevancy factors of input parameters were evaluated. The catalyst dosage and radiation time had the highest (32.6%), and the lowest (16%) relative contributions on the predicting of removal percentage of 2,4-DCP, respectively.