A new tool to predict the advanced oxidation process efficiency: Using machine learning methods to predict the degradation of organic pollutants with Fe-carbon catalyst as a sample

Herein, machine learning approaches were employed to predict the kinetic constant of the organic pollutant degradation process in a peroxymonosulfate environment with a typical Fe-carbon catalyst. After adjusting the hyperparameters and missing data imputation, an artificial neural network model was established, and the R2 value reached 0.9272. The model shows that catalyst dosage (12.4145%), pore volume (7.0642%), pollutant dosage (6.3571%), S value (5.3543%), and B value (4.2421%) of the linear solvation energy relation (LSER) model of pollutant are the top five important variables of all. Additionally, in the catalyst properties, pore volume, Fe-Nx content and graphitic N content have strongly positive effects, while specific surface area and oxygen content significantly inhibit the procedure. This work demonstrates a new optimization method for predicting the AOP efficiency, which further helps researchers recognize the process from a broad, comprehensive and innovative perspective.