Intelligent FA/FNA alternating strategy for nitrite-oxidizing bacteria inhibition: Data-driven prediction and process control.
Journal:
Journal of environmental management
Published Date:
Jun 1, 2025
Abstract
Alternating treatment with free ammonia (FA) and free nitrous acid (FNA) is an effective strategy to inhibit nitrite-oxidizing bacteria (NOB) in partial nitrification (PN) process. However, the current alternating treatment relies on manual assessment of nitrite accumulation rate (NAR), which poses challenges for automation. This study proposed an intelligent control system for automatic real-time switching between FA/FNA inhibition strategy and real-time control of inhibition concentration to realize stable NOB inhibition. By comparing the prediction performance of three models with different complexities for both classification and regression methods, support vector machine (SVM) was used to determine whether to alternate the strategies based on whether NAR was below 96 % and multilayer perceptron (MLP) was used for real-time control of FNA concentration by predicting the nitrite concentration. The high prediction accuracy of these two sub-models provides a solid foundation for the automatic control model of FA/FNA. Both models were trained and tested using an experimental dataset with manual alternating FA/FNA strategy over 180 days of a continuous flow PN reactor. After optimizing algorithms, the SVM had a 91.67 % classification accuracy, while the MLP showed an R of 0.96 and an RMSE of 53.16. During the real-time control of the intelligent control system, the SVM showed a classification accuracy of 97.5 % compared to actual measurements, and the R between the controlled FNA and the actual FNA is 0.83, with an RMSE of 0.04. The real-time operation demonstrates that the intelligent control system can promptly realize FA/FNA alternating and accurately control FNA concentration, maintaining NAR above 95.55 % while ammonium removal efficiency above 52.99 %. Compared to the current alternating treatment, the intelligent control strategy simplifies the manual operations and enables the automation of the FA/FNA alternating inhibition strategy, contributing to the stable and efficient operation of the PN process and promoting the application of PN/A process.
