Scholarlink

<p>The removal efficiency of Cr (VI) by chitosan (CS)-based composites under various working conditions can be accurately predicted using machine learning (ML) models trained on data from the literature. In this study, ensemble algorithms such as Extreme Gradient Boosting, Random Forest, and Adaptive Boost were employed for predictive modeling. Among these, the AdaBoost model demonstrated superior performance in forecasting the adsorption capacity of CS-based materials for Cr (VI) in aqueous solutions. Feature selection analysis identified initial Cr (VI) concentration, reaction time, adsorbent dosage, and solution pH as critical input parameters influencing adsorption capacity, with solution pH exerting the most significant impact (71%). The AdaBoost model emerged as the most suitable for predicting Cr (VI) adsorption, achieving robust performance metrics (R² = 0.830, MSE = 5.812, MAE = 0.008). To validate the model, a novel CS-based adsorbent (biochar-nanochitosan-zirconium (BC-nCS-Zr)) was tested experimentally, yielding results closely aligned with the Adaptive Boost predictions (R² = 0.825, RMSE = 7.406). This study highlights the potential of ML models in optimizing Cr (VI) removal processes using CS-based adsorbents. By providing an efficient alternative to costly and time-intensive experiments, it presents a promising pathway to reducing water pollution and improving environmental and public health outcomes.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine Learning-based Prediction and Experimental Validation of Cr (VI) Adsorption Capacity of Chitosan-based Composites

  • 期刊论文

  • 发表日期：

    2025年5月29日

查看全文
  • Fatemeh Yazdi,
  • Mohammad Sepehrian,
  • Mansoor Anbia

摘要

The removal efficiency of Cr (VI) by chitosan (CS)-based composites under various working conditions can be accurately predicted using machine learning (ML) models trained on data from the literature. In this study, ensemble algorithms such as Extreme Gradient Boosting, Random Forest, and Adaptive Boost were employed for predictive modeling. Among these, the AdaBoost model demonstrated superior performance in forecasting the adsorption capacity of CS-based materials for Cr (VI) in aqueous solutions. Feature selection analysis identified initial Cr (VI) concentration, reaction time, adsorbent dosage, and solution pH as critical input parameters influencing adsorption capacity, with solution pH exerting the most significant impact (71%). The AdaBoost model emerged as the most suitable for predicting Cr (VI) adsorption, achieving robust performance metrics (R² = 0.830, MSE = 5.812, MAE = 0.008). To validate the model, a novel CS-based adsorbent (biochar-nanochitosan-zirconium (BC-nCS-Zr)) was tested experimentally, yielding results closely aligned with the Adaptive Boost predictions (R² = 0.825, RMSE = 7.406). This study highlights the potential of ML models in optimizing Cr (VI) removal processes using CS-based adsorbents. By providing an efficient alternative to costly and time-intensive experiments, it presents a promising pathway to reducing water pollution and improving environmental and public health outcomes.