Study on preparation of KF/Al2O3 nano-composite catalyst via support vector regression

In order to investigate the influence of experimental condition on the catalytic effect of prepared KF/Al₂O₃ nano-composite catalyst in terms of the Knoevenagel reaction, support vector regression (SVR) combined with particle swarm optimization (PSO) algorithm is adopted for generating a numerical model to predict the yield of resultant 2-cyanide-3-phenyl ethylacrylate under different dosage of KF·2H₂O, reaction time, reaction temperature and calcination temperature. The comparison between SVR model and multivariable nonlinear regression (MNR) function reveal that the SVR model could predict the yield more accurate. Meanwhile, multifactor analysis based on the established SVR model describes the interactive influence on resultant yield among each experimental factor, providing a good support for KF/Al₂O₃ nano-composite catalyst fabrication. This study suggests that, as an efficient approach in data process and analysis, SVR could provide an effective reference for fabricating the nano catalyst with higher catalytic performance.