Modeling And Control Of Wastewater Treatment Process Based On Multi-Objective Multi-Gene Genetic Programming

Wastewater treatment process is a large-scale industrial process with complex reaction mechanism,strong nonlinearity,high coupling during variables and a lot of uncertain factors Meanwhile,it is a difficult and hot issue to keep the safety,smooth,long-term and efficiency operation during the wastewater treatment process.This dissertation aims to implement optimization control of wastewater treatment process.A wastewater treatment process model by stiff constant differential equations was established,and the solving method was studied From the view of the process optimization control,the sensitivity of wastewater treatment process was analyzed,and the sensitivity relationship between key operation variables and water components were established.In the end of this dissertation,based on the stiff constant differential equations model for wastewater treatment process,a modeling method based on Multi-Objective&Multi-Gene Genetic Programming(MO-MGGP)was proposed.Based on the simple analytical function expression of the MO-MGGP model,the linear parameter varying(LPV)state space model interpretation of the polynomial model was exploited,and then a predictive controller was designed,a nonlinear predictive was achievedThe main research works in this dissertation are summarized as followsBased on the model of BSM1 which was proposed by the International Water Association(IWA)and European cooperation COST,after studying the reaction mechanism and the technological process of the wastewater treatment process,a wastewater treatment process model by stiff constant differential equations was established.Because of the dynamic mechanism model with high dimension and coupling of variables,activated sludge systems exhibit stiff dynamics,it is important to find the solving method for stiff constant differential equations.And then,a simulation platform was established by the software of MATLAB.The simulation results show that the dynamic mechanism model can represent the characteristics of the wastewater treatment process,meanwhile the solving method can improve the accuracy and speedFrom the view of the system engineering and process control,the relationship between key operation variables and water components of bio-wastewater treatment processes were analyzed for the frist time.Through the open-loop,closed-loop curve and sensitivity analysis method,the impact law of different control variables on water components and the coupling between variables were discussed.This provides a theoretical basis for subsequent research on optimization controlWastewater treatment plants are nonlinear and high coupling systems subject to large perturbations in flow and load,together with uncertainties on the composition of the incoming wastewater.Nevertheless these plants have to be operated continuously,meeting stricter and stricter regulations.A data-driven modeling method based on MO-MGGP was proposed in this dissertation.Because of the strong and concise explanation ability of Multi-Gene Genetic Programming,and the balance of modeling accuracy and complexity based on Multi-Objective Programming,the simple and generalization dynamic model of dissolved oxygen concentration and nitrate nitrogen concentration in the wastewater treatment process was establishedRegarding the computational complexities associated with the control law calculation and poor performance of real-time control,based on the simple analytical function expression of the MO-MGGP model,LPV state space model interpretation of the polynomial model is exploited,and then a predictive controller is designed.The result of simulation proves that the predictive controller achieves good control effect.