Process modeling of total nitrogen removal in rotating biological contactor through artificial neural networks and Gaussian process regression

Efficient removal of total nitrogen (TN) from domestic wastewater is essential to prevent eutrophication and protect receiving water bodies. In this study, the performance of a rotating biological contactor (RBC) for TN removal was investigated under varying operational conditions using an integrated experimental and data-driven modeling framework. The effects of three key operating parameters—disk rotational speed, hydraulic retention time (HRT), and sludge retention time (SRT)—were evaluated using artificial neural networks (ANN) and rational quadratic Gaussian process regression (RQ-GPR). Both models demonstrated strong agreement with experimental observations; however, the ANN provided improved generalization, achieving a coefficient of determination (R²) of 0.978, compared to 0.972 for RQ-GPR on unseen test data. ANN effectively captured complex nonlinear relationships during training, whereas RQ-GPR exhibited superior robustness and predictive reliability across the validation and test datasets. The combined use of ANN and RQ-GPR provides a comprehensive framework for optimizing TN removal in RBC. It provides valuable guidance on the design and operation of energy-efficient, sustainable biological wastewater treatment.