Study On Control Of The Dissolved Oxygen Concentration In Biological Wastewater Treatment Process Based On The Fuzzy Adaptive Generalized Predictive Control Method

Water has been irreplaceable for human’s production and life. However, because of increasingly serious water pollution problems, the sustainability of water itself has been facing unprecedented challenges, and these have threatened human’s production and living in turn. On the basis of the existing wastewater treatment process, the control problem of dissolved oxygen(DO) concentration of wastewater treatment process will be focused in the paper, and the wastewater benchmark simulation model(BSM1) is taken as object of my study. The activated sludge wastewater treatment process A2/O is introduced in BSM1, which provides a standard simulation platform for the study and compare of different control schemes for wastewater treatment. Against to the DO concentration in the aerobic tank of ASM1, the model and control problems have been researched, and the experimental results show that the modeling accuracy and generalization capability, robustness and convergence of the control algorithm. Three aspects will be researched in this paper.(1) Firstly, because wastewater treatment is a process of nonlinear, large time-varying, large delay and serious interference, it is difficult to modeling. The generalized dynamic fuzzy neural network model method(GDFNN) will be used to predict the change of DO concentration. An elliptical basis function(EBF) is used as its fuzzy membership function in this method, and the width of it will be adjusted according to the importance of input variables. At the same time, the accuracy and generalization ability of DO concentration model will be improved by its features, such as learn online, self-organization and trim of rule. The simulation results show the accuracy and generalization capability of it.(2) Secondly, Based on the result CARIMA model of DO, generalized predictive controller(GPC) has been designed based on system output error, using the Diophantine equation to get control variables and realize rolling optimization and feedback correction, then the controller is applied on BSM1, the simulation result shows that the effectiveness of GPC with less interference.(3) Thirdly, an adaptive fuzzy generalized predictive control algorithm(AFGPC) is proposed in this paper for these limitations, which use the Lyapunov stability theorem to correct parameters of DO model directly, and then on the basis of this correction model, GPC is used to control DO system. AFGPC is an explicit adaptive control algorithm, which can achieve the stability control and better performance of tracking especially with large disturbance.