A Study on the Sedimentation Model and Neural Network Online Adaptive Control of a Benzoic Acid Imitated Wastewater Oxidation Process

A chemical oxidation method was applied to treat the wastewater in this study. Benzoic acid and water were mixed to imitate the wastewater with different values of chemical oxygen demand (COD). Hydrogen peroxide and ferrous chloride were both added to treat the wastewater. After the completion of oxidation, anion resins were added to coagulate and hence settle down the suspended particles. Therefore, the factors such as pH affecting the coagulation, flocculation, and sedimentation were discussed. A new model derived from system engineering was also proposed to describe the sedimentation process. Different adsorbents were used for the evaluation of the coagulation effect. In addition, a backpropagation neural (BPN) network was applied to the online adaptive control of the continuous water treatment system in this work. The COD of the treated effluent water was set to 90 ppm. The purpose of the control is to provide a minimum amount of reagents to reach a required COD of 90 ppm. The neural network structure was properly determined as 7-4-1; the only output node was the predicted H₂O₂ that would be added at the next control time. The network was a time-delayed BPN. The concentration of the added reagents is an important factor and was compared as well. The neural network controller was trained in a dynamic mode during the online operation. The data size for neural network online learning was properly determined experimentally.