Predictive Modeling and Feature Importance Analysis for Aeration Energy Optimization in Extended Aeration Wastewater Treatment Plants: Case Study Eyvan Wastewater Treatment Plant

The high energy demand of wastewater treatment plants (WWTPs), particularly for aeration, poses significant operational and economic challenges. This study presents a comprehensive machine learning approach for predicting energy consumption in aeration basins of extended aeration wastewater treatment plants. Using ۱۰ years of daily operational data, we developed Random Forest and XGBoost models to identify key parameters influencing energy usage. The analysis revealed that MLSS (Mixed Liquor Suspended Solids) is the most significant predictor, accounting for ۳۹.۲% of energy consumption variation, followed by COD (۲۷.۷%) and inflow rate (۱۳.۷%). The models achieved impressive prediction accuracy with R² scores of ۰.۶۵ (Random Forest) and ۰.۶۸ (XGBoost), demonstrating the potential for substantial energy optimization through data-driven operational adjustments. This study concludes that data-driven models coupled with feature importance analysis provide a powerful framework for plant operators to identify key leverage points for targeted energy optimization, leading to more cost-effective and sustainable wastewater treatment operations at the Eyvan plant and similar facilities.

کلیدواژه ها:

sewage treatment ، energy efficiency ، ecological balance ، biodiversity ، sustainability