AI-assisted predictive monitoring of process control parameters in a novel sequencing batch biofilm reactor for high-strength sewage treatment

Abstract This study presents a novel three-zone fibers-based sequencing batch biofilm reactor designed to achieve simultaneous nitrification, denitrification, and phosphorus removal for the treatment of high-strength sewage. Online process control sensors continuously monitored ORP, pH, and DO to provide real-time insight into biological phase transitions within the reactor. A sliding window-based linear regression model was developed and applied as an AI-assisted predictive monitoring framework, enabling accurate forecasting of process control parameters. The SBBR consistently achieved high removal efficiencies for organics and nutrients, producing effluent quality of BOD <3 mg/L, COD <35 mg/L, TN <5 mg/L, and TP ≈1 mg/L, meeting stringent regulatory standards. The AI predictive model demonstrated strong agreement with measured data for all three parameters (pH: R2 = 0.982, RMSE = 0.042; ORP: R2 = 0.996, RMSE = 7.951; DO: R2 = 0.918, RMSE = 0.173), confirming its reliability as a process monitoring tool for identifying biological phase endpoints in real-time. These findings establish the integration of AI-based predictive monitoring with advanced biofilm reactor technology as a practically viable and scientifically sound for high-strength sewage treatment.