Artificial Intelligence for Environmental Contaminant Removal: Cesium Ion Adsorption on Ti-MOF/rGO

Efficient removal of cesium ions from aqueous environments remains an important environmental challenge due to their high mobility and solubility. Here, an NH۲-MIL-۱۲۵(Ti)/reduced graphene oxide (rGO) hybrid composite is developed as an effective adsorbent for cesium removal from water. Operational parameters including pH, initial ion concentration, adsorbent dosage, and contact time are used as inputs to predict adsorption efficiency. Three regression-based algorithms-Random Forest, Multi-Layer Perceptron, and Extreme Gradient Boosting-are implemented and compared to evaluate their predictive performance. Results show that ensemble and kernel-based learning methods effectively capture the nonlinear relationships governing the adsorption process. Among the tested models, MLP demonstrates superior accuracy and robustness in predicting system behavior. Feature analysis further highlights the dominant influence of pH and initial concentration on removal efficiency. The study emphasizes the strength of data-driven modeling in understanding complex adsorption systems and provides a reliable predictive framework for optimizing process conditions in environmental applications.

کلیدواژه ها:

cesium removal ، titanium-based metal-organic frameworks /reduced graphene oxide adsorbent ، machine learning ، predictive modeling