Dynamic Optimization And Control Of The Aeration Process Based On A Simplified ASM1 Model

With the increasing requirements of environmental protection, optimal operation of the wastewater treatment plant is very important. Since aeration process is a major part of energy consumption in wastewater treatment process, in this paper, dynamic optimization and control is applied to the aeration process so the aeration energy consumption can be reduced.We introduce the ASM1 (activated sludge model No.1) with every detail. In order to research the energy consumption of aeration process, we developed a simplified model for aeration process based on ASM1 and use AMPL language for modeling, and simulating aeration process of the fifth pool of BSM1. The comparison results show that the model is correct. For the solution of the model, we discrete state variables and the control variables in finite element, change nonlinear differential equations of DAE system into nonlinear algebraic equations, call IPOPT solver to solve. The aeration system model laid the foundation for the research of different optimal control strategies.A minimum aeration energy consumption function as the objective function of the aeration optimal control problem was presented. We use the IPOPT solver to solve the optimal control problem under dynamic influent water conditions. The result shows that dynamic optimization of the aeration process was achieved.A minimum aeration energy consumption function as selected as part of the performance indicators and the aeration predict control model was established in AMPL. Also use the IPOPT solver to solve the model predict control problem under dynamic influent water conditions. Compared with the dissolved oxygen PID control, the effect of Model Predict Control is good and can save 3.2% of the energy consumption when use the appropriate parameters.A moving horizon estimation model based on aeration system model was presented in AMPL. Comparison of estimation results of different window length was made. The results of estimation are the best when the window length is 4. The results show that use the dynamic optimization method to solve MHE problem is successful.