Biomass-Derived Activated Carbon for Congo Red Dye Adsorption: Machine-Learning-Based Prediction and Comparative...

Area of Science:

• Environmental Chemistry
• Materials Science
• Computational Chemistry

Background:

• Congo Red (CR) is a carcinogenic azo dye prevalent in industrial wastewater.
• Effective removal of CR is crucial due to its persistence and environmental risks.
• Activated carbon from biomass offers a sustainable adsorbent material.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting CR dye adsorption onto Spathodea campanulata activated carbon (SCAC).
• To identify the most effective modeling approach for optimizing CR removal processes.
• To understand the influence of operational parameters on dye adsorption.

Main Methods:

• Utilized a dataset of 180 batch experiments with varying pH, dosage, initial CR concentration, contact time, and temperature.
• Constructed and compared four machine learning models: Support Vector Machine (SVM), Adaptive Neuro-Fuzzy Inference System (ANFIS), Artificial Neural Network (ANN), and Multiple Linear Regression (MLR).
• Performed global sensitivity analysis on the optimized ANFIS model to rank parameter importance.

Main Results:

• ANFIS (R² = 0.9722) and ANN (R² = 0.9710) exhibited superior performance in modeling CR adsorption compared to SVM (R² = 0.9597) and MLR (R² = 0.8667).
• MLR showed limited capability in capturing the nonlinear adsorption behavior.
• Sensitivity analysis indicated contact time and initial CR concentration as the most influential parameters for adsorption.

Conclusions:

• ANFIS provides a robust and accurate framework for modeling and optimizing Congo Red dye removal using SCAC.
• Machine learning approaches, particularly ANFIS, are valuable tools for predicting and enhancing wastewater treatment processes.
• SCAC shows significant potential for the efficient removal of carcinogenic azo dyes from industrial effluents.