Machine Learning-Based Prediction of Acidogenic Performance in Anaerobic Fermentation of Chemical-Biological Sewage Sludge
Municipal sludge anaerobic resource recovery efficiency in China lags behind developed countries. Widespread chemical phosphorus removal increases iron and aluminum salt precipitates in waste activated sludge, forming chemical-biological sludge that reduces acidogenic efficiency. This study identified key factors and developed a high-precision prediction model. Integrating literature and experimental data, acidogenic performance indicators under various conditions were compiled. Five machine learning models—Backpropagation Neural Network, Adaptive Neuro-Fuzzy Inference System, Support Vector Machine, K-Nearest Neighbors, and Random Forest—were systematically compared. Random Forest achieved the best predictive performance with a test set coefficient of determination (R²) of 0.9463, significantly outperforming others with minimal overfitting risk, demonstrating strong capability for high-dimensional, nonlinear, multi-factor coupled problems. Feature importance analysis revealed pH and Volatile Suspended Solids (VSS) as primary drivers, with aluminum salts exerting greater influence than iron salts. Engineering optimization should follow the pathway: 'adjust pH, stabilize organic matter, control aluminum salts'. This study provides an intelligent predictive tool and clarifies optimization directions, advancing precision and intelligent sludge treatment.