Prediction of Breakthrough Curve in FiXed Bed Adsorption Column using Artificial Intelligence

This paper proposes a new methodology for predicting the biochar Hydrogen Sulfide Adsorption Breakthrough Curve in the biogas desulfurization process. The proposed method can also be implemented for other adsorption processes using fixed bed columns. A Machine Learning model was devised and evaluated to predict the experimental breakthrough curve based on a developed biochar Hydrogen Sulfide adsorption dataset. The Gompertz function was implemented using two coefficients to express the experimental breakthrough curves in mathematical form. Also, various operation conditions were considered in developing the dataset. The eXtreem Gradient Boosting Regression was selected as an accurate model for predicting the biochar adsorption breakthrough curve. Moreover, Features Important analysis was implemented to derive valuable insight into the effect of features on the breakthrough curve. eXtreem Gradient Boosting Regression model could acquire an R۲ score of ۰.۹۹۶ in the training set and ۰.۹۷۴ and ۰.۸۵ in the testing and ۵-fold Cross-Validated sets, respectively, representing the model’s appropriate accuracy and robustness.

کلیدواژه ها:

Machine Learning ، Biogas Desulfurization ، Breakthrough Curve Prediction ، AdsorptionColumn.