基于机器学习的抗生素对厌氧氨氧化脱氮抑制效应分析OA

Anaerobic ammonium oxidation(anammox),as a novel nitrogen removal process,is widely used for treating high-ammonia nitrogen wastewater.However,antibiotics present in wastewater can inhibit microbial activity and reduce nitrogen removal efficiency.Investigating the mechanism by which antibiotics inhibit nitrogen removal efficiency has become key to improving the process.Therefore,this study developed predictive models for nitrogen removal efficiency under antibiotic inhibition using four nonlinear algorithms:Categorical Boosting(CatBoost),Extreme Gradient Boosting(XGB),Random Forest(RF),and K-Nearest Neighbors(KNN).All four machine learning models accurately predicted nitrogen removal efficiency(R2>0.95),with the XGB model showing better goodness-of-fit(adjusted R2adj=0.992)and the CatBoost model demonstrating superior predictive ability(external Q2ext=0.947).In addition,SHAP(SHapley Additive exPlanations)values were used to interpret the models,revealing that hydraulic retention time,influent nitrogen concentration,and antibiotic concentration are the main factors influencing nitrogen removal efficiency.This study also analyzed the mechanisms through which these factors affect nitrogen removal efficiency,providing a theoretical basis for improving nitrogen removal efficiency and optimizing the operational parameters of the anammox process.