Machine Learning-Assisted Analysis of Chloride-Enhanced BiOCl/DSA Anode for Synergistic Denitrification and Tetracycline Degradation: Performance, Mechanisms, and Product Regulation

Electrochemical co-treatment of nitrogen-containing and antibiotic wastewater is constrained by competitive oxygen evolution and insufficient chlorine evolution selectivity, while the resulting complex reaction network remains difficult to resolve. Therefore, a chloride-enhanced BiOCl/DSA anode coupled with a ZVI/Co3O4/GF cathode was constructed for deep denitrification and simultaneous tetracycline (TC) degradation. A fuzzy neural network (FNN) was further used to predict NH4+-N formation and conversion. The effects of initial NO3−-N and TC concentrations, KCl concentration, pH and cathode potential were systematically examined. At each tested KCl concentration, the BiOCl/DSA anode yielded a higher free residual chlorine concentration than the unmodified DSA anode. Relative to the commercial DSA anode, the BiOCl/DSA anode increased apparent dissolved inorganic nitrogen (DIN) removal from 60.17% to 71.89%, decreased NH4+-N accumulation from 44.54 to 28.48 mg/L, and increased TC removal from 74.51% to 100%. Complete DIN removal was achieved at 1000 mg/L KCl and a cathode potential of −1.4 V versus Ag/AgCl, while TC was efficiently degraded within 80 min. The BiOCl layer may promote chloride adsorption and chlorine-radical generation, thereby facilitating NH4+ oxidation and TC degradation. Among the four models compared, FNN showed the best internal fitting performance under the random 80:20 data split (test-set R2 = 0.9507). Given the limited size of the dataset, the model was used only for exploratory analysis within the range covered by the experimental data, and its generalizability to new operating conditions remains to be established using independent validation data.