Control Model Establishment Of SBR System Using Neural Network & Validated By Experiment

The simulation and control model establishment of sewage treatment system is a new research and development direction. The paper used BP artificial neural network for its ability of function impending and optimization to figure out the most optimal aeration volume and aeration time during controlling SBR system. Network input variables and output variables were confirmed by finding factors affecting aeration volume and aeration time based upon ancestors' research work. Training swatches and verifying swatches were obtained by experiments. Through training the network and comparing results, the structure of network was settled to 3-4-2, at the same time, values of network parameters were determined. Based upon the processing of network computing, the expression of control model on SBR system was done and values of parameters were obtained. Also verifying swatches and the verifying experiment verified the feasibility of the model. In order to evaluate the control model, sensitivity analysis was applied to effect of influent COD concentration, initial DO value, and initial MLSS concentration respectively upon aeration volume and aeration time. The paper offered a new method to construct control model of SBR system.