Improving nitrogen removal using a fuzzy neural network-based control system in the anoxic/oxic process

Area of Science:

• Environmental Engineering
• Wastewater Treatment Technologies
• Artificial Intelligence in Environmental Science

Background:

• Biological wastewater treatment processes are complex and difficult to control, particularly nitrogen removal.
• Existing methods struggle with inherent uncertainties and the need for cost-effective operations.

Purpose of the Study:

• To develop an integrated neural-fuzzy control system for enhanced nitrogen removal in anoxic/oxic (A/O) wastewater treatment.
• To improve the cost-effectiveness and control performance of biological nitrogen removal processes.

Main Methods:

• Developed a fuzzy neural network (FNN) predicted model for forecasting nitrate concentration.
• Implemented an FNN controller to manage nitrate recirculation flow.
• Embedded self-learning capabilities within the FNN for improved rule extraction and network performance.

Main Results:

• Achieved reasonable forecasting and control performances for nitrogen removal.
• Demonstrated significant reductions in effluent Chemical Oxygen Demand (COD) by approximately 14%.
• Reduced effluent Total Nitrogen (TN) by about 10.5% and operational costs by 17%.

Conclusions:

• The integrated neural-fuzzy control system effectively addresses the challenges of nitrogen removal in wastewater treatment.
• The developed system offers a cost-effective solution with improved effluent quality and operational efficiency.