Exploring sludge yield patterns through interpretable machine learning models in China's municipal wastewater treatment plants

Abstract

Sludge management remains a challenge for municipal wastewater treatment plants (MWWTPs). In this study, we use machine learning models to predict sludge yield and employ interpretable methods to highlight the driving factors. We analyze over 27,000 data entries of monthly plant-level operational details to predict the sludge yield for 177 MWWTPs in 11 cities throughout China. Evaluated by multiple statistical indicators including Coefficient of Determination (R2), Mean Absolute Error (MAE), Normalized Mean Absolute Error (NMAE), Mean Square Error (MSE) and Root Mean Square Error (RMSE), the machine learning model's performance proves superior to empirical estimation. Interpretative analysis reveals that pollutant removal quantities exert a more substantial influence on sludge yield than influent pollutant concentrations. The sludge yield becomes increasingly sensitive to wastewater quality when effluent discharge standards rise. The integration of interpretable machine learning models expands the research scope to a more holistic perspective, catalyzing interdisciplinary collaboration and novel insights.