CatBoost-Based Multi-Objective Optimization of the Heterogeneous Electro-Fenton Process for Tetracycline Removal and Energy Efficiency

Purpose: This study presents a data-driven approach for optimizing the heterogeneous electro-Fenton process applied to pharmaceutical wastewater treatment using the CatBoost machine learning model combined with the Non-dominated Sorting Genetic Algorithm II (NSGA-II). Methods: Experimental data were obtained from a ۱ L electrochemical reactor using MIL-۱۰۰(Fe) as a heterogeneous catalyst and persulfate as an oxidant. Results: CatBoost was trained to predict tetracycline (TC) removal efficiency and electrical energy consumption, achieving reliable performance with RMSE values of ۸.۵۱% and ۵۶.۶۳ kWh/kg, MAE values of ۷.۶۳% and ۵۰.۹۶ kWh/kg, and R² values of ۰.۶۷ and ۰.۷۹ for TC removal and energy consumption, respectively. The optimization, with weighting factors of ۰.۷ for TC removal and ۰.۳ for energy minimization, yielded an optimal trade-off point of ۷۹.۹۴% TC removal and ۲۰۹.۸۵ kWh/kg energy consumption, with an overall weighted score of ۰.۷۱۶. Conclusion: The results confirm the potential of CatBoost-based modeling integrated with evolutionary optimization to enhance environmental performance and energy efficiency in advanced oxidation processes.

کلیدواژه ها:

Electro-Fenton ، tetracycline ، MIL-۱۰۰(Fe) ، CatBoost ، multi-objective optimization ، NSGA-II