Prediction of CO2 adsorption on different activated carbons by hybrid group method of data-handling networks and LSSVM

Abstract

ABSTRACTCO2 is a ubiquitous species that has received much attention recently. The adsorption of CO2 by means of activated carbons is a well-tried technology that can be used on a large scale. Improvements of the prediction models with more accurate results and lower error are necessary for future development of the projects and the economic dispatch sector. The least square support vector machine, a relatively unexplored neural network known as group method of data handling (GMDH), were implemented to forecast the CO2 adsorption on different activated carbons. This work aims to provide new methods to predict the adsorption equilibrium of pure CO2 on a set of commercial activated carbons and to express it regarding textural properties such as Brunauer–Emmett–Teller (BET) surface area, total pore volume, and micropore volume. Results indicated that the utilized models are very accurate in predicting CO2 adsorption on different activated carbons. Comparison of the outcomes of the two models shows that the GMDH model is more accurate with R2 and mean squared error values of 0.8915 and 0.0001425, respectively.