基于ANN和NSGA-Ⅱ的磷钨酸/SiO2催化酯交换合成4-丙烯酸羟丁酯工艺优化OA

Using an impregnation method,phosphotungstic acid(PTA)/SiO2 catalysts with varying PTA loadings were prepared with SiO2 carriers synthesized via the Stöber method.The structural and acidic properties of the catalysts were characterized using methods including XRD,FTIR,NH3-TPD,and TG.A single factor experiment was conducted with the process parameters for the transesterification of 1,4-butanediol(BDO)and methyl acrylate to produce 4-hydroxybutyl acrylate(4-HBA).A Bayesian optimized BP neural network(ANN)model was constructed using Matlab.BDO conversion and 4-HBA selectivity under various reaction parameters were predicted,including reaction temperature,reaction time,catalyst amount,PTA loading and ester/alcohol molar ratio.The second-generation non-dominated sorting genetic(NSGA-Ⅱ)algorithm was used to optimize the process parameters.According to the predicted results,the prediction values of the optimized process parameters under experimental conditions exhibit good agreement with the actual experimental values,which demonstrates that the constructed ANN model with NSGA-Ⅱ algorithm can accurately describe and optimize the transesterification reaction system.