| Activated sludge biological wastewater treatment is one of the most popular aerobic biological treatment technologies. The principle of purification is to introduce contact with bacteria (cells) or other microorganisms, which feed on the organic materials in the wastewater, thereby these organic materials will be transformed into cellular mass and some inorganics. In the light of the inherent complexity of the mechanism of wastewater treatment processes providing with high nonlinearity, strong coupling, uncertainty and large variations in the influent wastewater flow rate, concentration and composition, investigating an useful computer simulations, mathematical modelling and control method to improve the degree of automation and treatment effect in wastewater treatment processes, has great practical significance in relieving the shortage of water resources in our country. In this paper, a computer simulation platform is built, a fuzzy modelling algorithm and a model predictive control (MPC) based on the obtained fuzzy model are studied on the basis of the Benchmark Simulation Model N0.1 (BSM1).Firstly, the Benchmark of IWAQ is studied in depth as well as the activated sludge model No.1 (ASM1) and the double exponential secondary settlement tank model used in the Benchmark. Then, the simulation platform is acquired using Matlab, the reliability of the simulation model is tested via comparing the effluent data of simulation model with the standard data provided by the BSM1, which offers a necessary foundation for next work.Secondly, considering the high nonlinearity and strong coupling of most activated sludge process modells, a new approach to predict the concentration of the chemical oxygen demand (COD), a common criterion of the outlet water in most wastewater treatment plants, is discussed. At the beginning, the structure identification of fuzzy rules is studied based upon the mechanism in the ASM1. Further, this method combines the well-known fuzzy c-means cluster algorithm and the least squares method to partition the fuzzy space of input variables and identify the consequent parameters. The obtained model via the proposed method has simple structure, can be understood easily, keep the meanings of variables and can make better use of the theory about linear systems. At last, the promise of the approach is verified by comparing the predicted dynamics with experimental measurements obtained from the BSM1.Finally, the predictive control which has great tolerance is chosed to get over the disturbance and the error between predictive model and practical plant. Moreover, considering the nonlinearity of activated sludge process, the nonlinear model predictive control is also studied. Specially, fuzzy predictive controller whose predictive model is derived by the proposed method in this thesis is used successfully. These strategies are described in details and the simulation results are compared, which show that better dynamic response, steadier output, less overshoot are given by the predictive control strategies, particularly, the nonlinear fuzzy predictive control strategies, among DMC, fuzzy predictive control and traditional PID control. |
PDF Full Text Request