Mechanism of purification of low-pollution river water using a modified biological contact oxidation process and artificial neural network modeling

Abstract A modified biological contact oxidation (MBCO) process was adopted to treat low-pollution river water, mainly contaminated with effluent from a sewage treatment plant, containing nitrate nitrogen as the main form of nitrogen. The effects of varying carbon sources, carbon-to-nitrogen (C/N) ratios and hydraulic retention times on the effectiveness of river water treatment were investigated, as well as the mechanism of pollutant removal. An artificial neural network (ANN) model optimized using a genetic algorithm was established to predict the effluent water quality after MBCO treatment. Results showed that when sodium acetate was applied as the carbon source, the MBCO reactor denitrification effect was optimal, with the denitrification efficiency increasing in accordance with increasing C/N ratio. When sodium acetate was applied as the carbon source at a C/N ratio of 4.5, the relative abundances of Anammox and denitrifying bacteria in the MBCO reactor were highest. These results indicate that these conditions were most favorable for the reproduction of Anammox and denitrifying bacteria, accounting for the improved denitrification effect, achieving a total nitrogen removal efficiency of ≥86.8%. A genetic algorithm optimized ANN model was applied, which accurately predicted the effluent pollutant concentration of low-pollution river water treated using the MBCO process. The MBCO process provides an effective alternative for the treatment of low-pollution and low C/N ratio water.