Hybrid machine learning–NSGA-II framework for multi-objective optimization of anaerobic digestion in wastewater treatment

Abstract Anaerobic digestion (AD) is a key process for sludge treatment, volume reduction, and energy recovery in wastewater treatment plants (WWTPs). Therefore, the use of artificial intelligence methods in ED is recommended for better sludge and wastewater management. With the development of artificial intelligence, this study investigates the laboratory-scale AD system using an estimation-optimization framework based on input sludge characteristics. Three machine learning (ML) techniques - random forest (RF), support vector regression (SVR), and multilayer perceptron (MLP) were applied - to estimate biogas yield (biogas), total solids in the effluent (TS out ), and degritting cost (cost) under varying sludge conditions. Model performance was assessed using the coefficient of determination (R²), mean square error (MSE), and mean absolute relative error (MARE). The results showed strong correlations between input data and output parameters, with the best model (SVR) for biogas achieving an R² of 0.935, MSE of 0.00111, and MARE of 5.2959%. Subsequently, multi-objective optimization was conducted by connecting the best-performing ML models with the Non-dominated Sorting Genetic Algorithm II (NSGA-II). Then, the beam responses were evaluated using the TOPSIS method with different weightings. The optimization results indicate that, by examining the decision variable across three objectives, the minimum achievable values of TS out and cost are 0.0801 kg/d and 7.5170 US$, respectively, while the maximum predicted biogas is 0.4597 m 3 /Kg VS.