A Robust Method to Predict Equilibrium and Kinetics of Sulfur and Nitrogen Compounds Adsorption from Liquid Fuel on Mesoporous Material

This study presents a robust and rigorous method based on intelligent models, namely radial basis function networks optimized by particle swarm optimization (PSO-RBF), multilayer perceptron neural networks (MLP-NNs), and adaptive neuro-fuzzy inference system optimized by particle swarm optimization methods (PSO-ANFIS), for predicting the equilibrium and kinetics of the adsorption of sulfur and nitrogen containing compounds from a liquid hydrocarbon model fuel on mesoporous materials. All the models were evaluated by the statistical and graphical methods. The predictions of the models were also compared with different kinetics and equilibrium models. The results showed that although all the models lead to accurate results, the PSO-ANFIS model represented the most reliable and dependable predictions with the correlation coefficient (R2) of 0.99992 and average absolute relative deviation (AARD) of 0.039%. The developed models are also able to predict the experimental data with better precision and reliability compared to literature models.