ANN and GPR Modeling of Multipollutant Removal in Rotating Biological Contactor

Effective treatment of domestic wastewater requires biological processes capable of maintaining stable performance under varying operational conditions. This study applies artificial neural network (ANN) and Gaussian process regression (GPR) to model and predict multipollutant removal performance in a lab‐scale rotating biological contactor (RBC). The RBC performance was evaluated for the removal of chemical oxygen demand (COD), ammonium‐N, and turbidity under different hydraulic retention time (HRT), sludge retention time (SRT), and disk rotational speed. ANN model demonstrated strong predictive agreement with experimental observations, effectively capturing nonlinear relationships between operational parameters and removal efficiencies. GPR provided accurate predictions together with quantitative uncertainty estimates, offering additional insight into model confidence and sensitivity across operating parameters. The combined ANN–GPR framework highlights the trade‐off between predictive accuracy and uncertainty‐aware interpretation in data‐driven modeling of the RBC bioreactor. The findings enhance understanding of parameter interactions in RBC and support the use of machine learning–based tools for performance analysis and decision support for wastewater treatment.