미생물 연료전지 기반 자가구동형 전기분해 소독 시스템 개발 및 유리염소 생성의 AI 예측 = Development of Microbial fuel cell-Based Self-Powered Electrolytic Disinfection system and AI Prediction of Free chlorine Generation

다국어 초록 (Multilingual Abstract)

Growing concerns over pathogenic microorganisms in wastewater have increased the demand for energy-efficient and chemically safe disinfection technologies. This study developed a self-powered electrolysis system by integrating a microbial fuel cell (MFC) to generate free chlorine without external electricity. The parallel-connected MFC stack provided sufficient voltage to stably drive electrolysis, producing up to approximately 11 mg/L of free chlorine under 3% NaCl conditions. This allowed rapid inactivation of bacteria, achieving more than 5-log reductions of Escherichia coli and Listeria monocytogenes within 2 minutes. In contrast, the non-enveloped MS2 bacteriophage exhibited strong resistance, showing less than 1-log inactivation after 15 minutes, while supplementary power-supply experiments confirmed that viral removal requires a current higher than the power generated by the designed system. Machine-learning models were further developed to predict free chlorine concentration and pH from electrochemical parameters. Random Forest and ExtraTrees achieved high predictive accuracy (R² > 0.9), and transfer-learning approaches applied to CatBoost and XGBoost improved model stability by leveraging patterns learned from power-supply data. Overall, this study demonstrates the feasibility of MFC-driven electrolysis as a sustainable approach for on-site disinfectant production, and the developed predictive framework provides a foundation for autonomous operation and scale-up of MFC-based water treatment systems.

목차 (Table of Contents)

• 1. Introduction 1
• 1.1. General Background 1
• 1.2. Microbial fuel cell and Basic Principles 4
• 1.3. Literature survey and aim 5
• 2. Experimental Section 8
• 2.1. MFC Reactor Setup and Operation 8
• 2.2. MFC-PMS–Based Free Chlorine Generation System 12
• 2.3. Microbial Inactivation Using the MFC-PMS–Based Free Chlorine Generation System 12
• 2.4. Electrochemical Characterization 16
• 2.5.Chemical Characterization 16
• 2.6. Microbial Characterization 17
• 2.7. Experimental Data Collection 17
• 2.8. Data Preprocessing 18
• 2.9. Development and Evaluation of Machine Learning Models 20
• 3. Results and Discussion 22
• 3.1. Electrochemical Characteristics of Microbial Fuel Cells for Disinfection Experiments 22
• 3.2. Operation of the MFC-PMS Free–Based Free Chlorine Generation System 25
• 3.3. Free Chlorine Production Driven by MFC-Integrated Electrolysis System 28
• 3.4. Bacterial Removal Kinetics in Electrolytic Disinfection Systems Operated Under Variable Conditions 31
• 3.5. Viral Removal Kinetics in Electrolytic Disinfection Systems Operated Under Variable Conditions 34
• 3.6. Viral Inactivation Kinetics Under Power Supply–Driven Electrolytic Disinfection 37
• 3.7. Electrochemical Characteristics of Microbial Fuel Cells for Prediction Experiments 39
• 3.8. Comparison of Electrolytic Performance Between PS and MFC-PMS Systems Under Varying NaCl Concentrations 42
• 3.9. Correlation Structure and Principal Component Patterns of PS and MFC-PMS Datasets 45
• 3.10 .Performance of Ridge, Random Forest, and ExtraTrees Models Without Transfer Learning 48
• 3.11. Performance Evaluation of Fine-Tuned CatBoost and XGBoost Models After Transfer Learning 51
• 4. Conclusions 54
• 5. References 55