Neural network and genetic algorithm for modeling and optimization of effective parameters on synthesized ZSM-5 particle size

Abstract

The aim of this study is to design a nonlinear model based on neural network for estimation of ZSM-5 particle size. The experimental data were gathered from the literature. The main preparation variables affecting ZSM-5 particle size are SiO2/Al2O3, template/SiO2, H2O/SiO2, and SiO2/Na2O ratios, crystallization time, and temperature, which are selected as input variables of the neural model. The results indicate that the designed model can exactly estimate the effects of six input parameters on the ZSM-5 particle size. An optimization paradigm based on genetic algorithm is employed to determine the minimum value of ZSM-5 particle size and the corresponding operating conditions. The optimized gel composition is as follows: SiO2/Al2O3=113.6, template/SiO2=0.32, H2O/SiO2=42.10, SiO2/Na2O 73.21 and the optimal temperature and crystallization time are 443K and 98h, respectively.