Optimized strategies for achieving nitrosative accumulation-based on a comprehensive understanding of nitrogen form transformation

The nitrosation process has demonstrated significant advantages, including energy savings and reduced organic matter consumption during nitrogen removal from wastewater. However, a comprehensive understanding of nitrosation under the influence of multiple factors is still lacking. This study aimed to clarify the priority and boundary conditions of limiting factors affecting nitrogen morphology and evaluate the multiple effects of influent substrate, reactor type, and operating parameters on the nitrification process. Four datasets (SBR-C, SBR-IC, ALR-low, and ALR-high) were established based on different reactor types, carbon source types and influent NH₄⁺-N concentrations. Several machine learning models were used to predict the morphological distribution of nitrogen compounds in the biological treatment effluent. BGDT was identified as the optimal model, which was then used to inversely predict the boundary conditions of operating parameters based on influent conditions. Correlation and SHAP analyses indicated that the most critical regulatory factors in the SBR-C, SBR-IC, ALR-low, and ALR-high systems were pH, C/N, influent NH₄⁺-N concentration, and DO/TAN, respectively. Lowering the influent C/N before nitrification in the aerobic unit was more conducive to NO₂⁻-N and AOB accumulation. Theoretical complete nitrosation was achieved when the C/N of the SBR-C and the SBR-IC were 1.48 and 1.14, respectively. Overall, maintaining weak bases, short hydraulic retention time (HRT), and long sludge retention time (SRT) in the reactor promoted the establishment of anaerobic ammonia oxidation conditions via nitrite nitrogen accumulation. In summary, machine learning can unify and optimize the regulation of material metrology, environmental conditions and microbial functions to change the carbon source consumption pattern, thus driving the development of nitrogen removal technologies with lower consumption, higher efficiency, and less discharge.