Research On Ozone Automatic Dosing System Of Wastewater Treatment Plant

With the development of economy and the accelerated process of urbanization, water pollution becomes more serious. The disinfection process of water and wastewater treatment plant becomes more important. At present, China’s urban wastewater treatment method generally use the traditional process-chlorine oxide process, but with the increasing of water pollution, the dosage of disinfectant also shows an increasing trend. However, chlorination process produces many carcinogenic, teratogenic, mutagenic substances. Ozone has disinfect effect and almost no secondary pollution, recent years, ozone has been widely used in wastewater treatment. Throughout the ozone disinfection process, the dosage of ozone determines the effect of disinfect and project operating costs. So the precise control of ozone dosage has practical significance.This thesis studies the ozone dosing control strategy for wastewater treatment plant. Using an on-line dissolved ozone analyzer to detect the concentration of the dissolved ozone in the water, which is the indicator, a closed loop is set up to regulate the ozone dosage. In order to improve the accuracy and robustness of the system, a cascade control system is set up in this thesis. In order to overcome the delay of chemical reaction and the instrumentation, this thesis use a Smith predictor control system. The Smith’s model uses second-order model which is an engineering approximation model, through actual data derived from engineering model. This model was designed as a controlled object with a Smith predictor cascade control system. The parameters of this model have a certain relation with water quality and temperature. In a strong interference case, controlled object model may change, may deteriorate the accuracy of the Smith predictor results. In view of this situation, this thesis choose predictive control algorithm which doesn’t demand accurate model of the controlled object. Based on a large number of field data, this thesis use the BP neural network for nonlinear system identification, to identify the model as a post-prediction model, then design a predictive control of ozone dosing control system based on neural network.Compare with the Smith predictor which needs a linearized plant model, the non-linear neural network prediction model of ozone based on the large number of field data is more reasonable with system’s characteristics such as non-linear, time-varying. This system has more accuracy. Simulation results show that the predictive controller has short adjusting time, small overshoot, and good anti-disturbance capability. It’s helpful to improve the accuracy of dosing of ozone.