Ji-Yeon Kim

Membrane fouling remains a major challenge in full-scale membrane bioreactor (MBR) systems, reducing operational efficiency and increasing maintenance needs. This study introduces a predictive and analytic framework for membrane fouling by integrating artificial intelligence (AI)-driven feature engineering and explainable AI (XAI) using real-world data from an MBR treating food processing wastewater. The framework refines the target parameter to specific flux (flux/transmembrane pressure (TMP)), incorporates chemical oxygen demand (COD) removal efficiency to reflect biological performance, and applies a moving average function to capture temporal fouling dynamics. Among tested models, CatBoost achieved the highest predictive accuracy (R² = 0.8374), outperforming traditional statistical and other machine learning models. XAI analysis identified the food-to-microorganism (F/M) ratio and mixed liquor suspended solids (MLSSs) as the most influential variables affecting fouling. This robust and interpretable approach enables proactive fouling prediction and supports informed decision making in practical MBR operations, even with limited data. The methodology establishes a foundation for future integration with real-time monitoring and adaptive control, contributing to more sustainable and efficient membrane-based wastewater treatment operations. However, this study is based on data from a single full-scale MBR treating food processing wastewater and lacks severe fouling or cleaning events, so further validation with diverse datasets is needed to confirm broader applicability.

This study investigates membrane fouling control in a submerged anaerobic ceramic membrane bioreactor (AnCMBR) treating high-strength food wastewater (chemical oxygen demand (COD): 10–30 g/L). A hybrid strategy combining mechanical cleaning via a moving rubber blade (MRB) (termed anaerobic ceramic blade MBR (AnCBMBR)) with intermittent salt-assisted backwash (SAB) was tested to manage transmembrane pressure (TMP) and sustain treatment performance. During more than 300 days of field operation, MRB alone maintained stable TMP below 0.15 kg_f/cm² without backwashing, achieving more than 90% COD removal at a very short hydraulic retention time (HRT) of 1–2 days. Introducing intermittent SAB further stabilized operations and enhanced total phosphorus (T-P) removal by facilitating struvite formation through the interaction of MgCl₂ and phosphorus in the reactor. The AnCBMBR system demonstrated reliable, long-term fouling control and treatment efficiency, even under high organic loads, proving its viability for small-scale facilities managing concentrated food wastewater. This study advances practical strategies for sustainable anaerobic MBR operation under challenging industrial conditions.

The prediction of the chemical oxygen demand (COD) and total nitrogen (TN) in integrated anaerobic–anoxic–oxic (A2O) and anoxic–oxic (AO) processes (i.e., A2O+AO process) was achieved using a dynamic ensemble model that reflects the dynamics of wastewater treatment plants (WWTPs). This model effectively captures the variability in the influent characteristics and fluctuations within each reactor of the A2O+AO process. By employing a time-lag approach based on the hydraulic retention time (HRT), artificial intelligence (AI) selects suitable input (i.e., pH, temperature, total dissolved solid (TDS), NH₃-N, and NO₃-N) and output (COD and TN) data pairs for training, minimizing the error between predicted and observed values. Data collected over two years from the actual A2O+AO process were utilized. The ensemble model adopted machine learning-based XGBoost for COD and TN predictions. The dynamic ensemble model outperformed the static ensemble model, with the mean absolute percentage error (MAPE) for the COD ranging from 9.5% to 15.2%, compared to the static ensemble model’s range of 11.4% to 16.9%. For the TN, the dynamic model’s errors ranged from 9.4% to 15.5%, while the static model showed lower errors in specific reactors, particularly in the anoxic and oxic stages due to their stable characteristics. These results indicate that the dynamic ensemble model is suitable for predicting water quality in WWTPs, especially as variability may increase due to external environmental factors in the future.

Wastewater treatment plants (WWTPs) generate useful data, but effectively utilizing these data remains a challenge. This study developed novel ensemble tree-based models to enhance real-time predictions of chemical oxygen demand (COD) and total nitrogen (TN) concentrations, which are difficult to monitor directly. The effectiveness of these models, particularly the Voting Regressor, was demonstrated by achieving excellent predictive performance even with the small, volatile, and interconnected datasets typical of WWTP scenarios. By utilizing real-time sensor data from the anaerobic–anoxic–oxic (A2O) process, the model successfully predicted COD concentrations with an R² of 0.7722 and TN concentrations with an R² of 0.9282. In addition, a novel approach was proposed to assess A2O process performance by analyzing the correlation between the predicted C/N ratio and the removal efficiencies of COD and TN. During a one and a half year monitoring period, the predicted C/N ratio accurately reflected changes in COD and TN removal efficiencies across the different A2O bioreactors. The results provide real-time COD and TN predictions and a method for assessing A2O process performance based on the C/N ratio, which can significantly aid in the operation and maintenance of biological wastewater treatment processes.