Data-Driven Recursive Kinetic Modeling for Fenton Reaction

Area of Science:

• Environmental Chemistry
• Water Treatment Technologies
• Computational Chemistry

Background:

• The Fenton reaction is a key advanced oxidation process for degrading persistent organic pollutants in water.
• Accurate kinetic modeling of Fenton reactions is difficult due to complex mechanisms and sensitivity to operational variables.

Purpose of the Study:

• To develop a novel machine learning framework for predicting Fenton reaction kinetics.
• To establish a data-driven approach for kinetic analysis and treatment optimization.

Main Methods:

• Developed a multiple estimation recursive machine learning (MERML) framework.
• Trained MERML on experimental data from Fenton reactions of 12 phenolic compounds.
• Utilized MERML to predict kinetic profiles from initial reaction conditions.

Main Results:

• MERML accurately predicted kinetic profiles without prior mechanistic knowledge.
• The framework demonstrated superior accuracy, few-shot learning, robustness, and interpretability.
• MERML enabled data-driven kinetic analysis, including optimization and variable influence.

Conclusions:

• MERML offers a powerful, data-driven tool for modeling pollutant degradation in Fenton systems.
• This approach enhances water purification treatment optimization and kinetic analysis.
• Provides a novel computational framework for environmental reaction systems.