Predicting carbon dioxide adsorption capacity on 13X and 5A zeolite using artificial neural network modeling
عنوان دوره: 1
1دانشجوی دکترای مهندسی شیمی/ دانشکده مهندسی شیمی، دانشگاه تهران، تهران، ایران
3university of Tehran
4گروه آموزشی مهندسی شیمی، پردیس مرکزی دانشکده فنی، دانشگاه تهران، تهران، ایران
5University Of Tehran
In this study, artificial neural network modeling was used for the prediction of carbon dioxide adsorption capacity on two zeolite adsorbents 13X and 5A. Temperature and pressure were used as system inputs and adsorption capacity as output. To determine the network training algorithm, the optimal transfer functions in the hidden and output layers, and the optimal neuron, the coefficient of the determination and the root mean square error were calculated. After modeling the experimental data, the Levenberg–Marquardt back-propagation algorithm (BP) was used to train the network in both zeolites. The optimal number of neurons for both 13X and 5A zeolites were 10. Finally, the results of artificial neural network modeling and the Toth model, obtained by Wang, were compared. R2, coefficient of determination for artificial neural network, and Toth have obtained for 13X, 0.9969, and 0.9918 and 0.9941 and 0.9923 for 5A adsorption, respectively, which shows the high accuracy of the artificial neural network compared to the Toth model.