Research On Sulfide Adsorption And Separation Of Amorphous Nanoporous Materials Based On Monte Carlo And Ensemble Learning Algorithms

H₂S and SO₂ are common harmful gases in the environment and industry,and studying the adsorption separation characteristics of sulfides can help protect the environment and improve energy efficiency.Amorphous nanoporous materials have high porosity and specific surface area,showing superior adsorption performance.In order to quickly select the appropriate adsorbent from a large number of amorphous nanoporous materials,molecular simulation is used to study the adsorption and separation characteristics of amorphous nanoporous materials,and the influencing factors and performance prediction of gas adsorption of amorphous nanoporous materials are studied by ensemble learning algorithm.The giant regular Monte Carlo algorithm was used to simulate the absolute adsorption of amorphous nanoporous materials on H₂S and SO₂one-component gases at a temperature of 303 K,respectively,and two sets of datasets were obtained.Through analysis,it is found that the absolute adsorption capacity is closely related to pressure and material characteristics.Among them,density was negatively correlated with absolute adsorption capacity,other material characteristics were positively correlated with absolute adsorption capacity,and porosity and specific surface area had the most significant effects on absolute adsorption capacity.The adsorption separation characteristics of CH₄-H₂S and CO₂-SO₂binary mixed gases in 116 amorphous nanoporous materials were simulated by using the giant regular Monte Carlo algorithm,and the adsorption separation characteristics of some materials were predicted using the ideal adsorption solution theory,and the corresponding selectivities were obtained.Through the analysis of data,it is found that the ideal adsorption solution theory can accurately predict the absolute adsorption amount,and can predict the abnormal phenomenon in the adsorption process.Models such as Random Forest,Gradient Boosting Decision Tree,e Xtreme Gradient Boosting Tree and Categorical Boosting are used to construct integrated algorithms to realize the training and prediction of absolute adsorption,and by comparing the performance indicators of the four models,it is found that the extreme gradient boosting tree has good prediction performance.The application of parallel coordinate maps to the field of amorphous nanoporous material screening can provide some guidance for the work in the field of material synthesis while finding a predictive model with good performance.