Modeling And Optimal Control Of SCR Denitrification System For Thermal Power Unit

With the continuous increase of the proportion of wind power,solar power and other new energy power generation,the requirements of power grid for large-scale deep load regulation of thermal power units have further increased.The rapid variable load operation of the unit will cause frequent fluctuations in the combustion process,which will result in large changes of the NOx concentration in the flue-gas.SCR flue-gas denitration is the most widely used denitration technology for coal-fired boilers,but its production process is seriously affected by the flue gas parameters of the boiler outlet.The traditional PID control strategy widely used at the production site cannot adapt to the large-scale rapid variable load operation of the unit.In order to ensure the NOx emission standard of the SCR outlet and the economic operation of the SCR denitration system,it is necessary to research an intelligent control strategy with a certain load adaptability,so the amount of ammonia injection can be precisely controlled.An accurate model of the system is a necessary premise for implementing optimal control of SCR system.Based on the analysis of the denitration reaction mechanism of SCR system,a simplified mechanism model was established,and the model parameters were identified by using particle swarm optimization.Although the mechanism model has good explanatory and dynamic characteristics,as the unit operating conditions vary widely,the calculation results of the mechanism model always deviate from the actual situation.In order to solve this problem,the data-driven model of the SCR denitration system is established based on LSSVM,and then the hybrid dynamic model is developed via arranging the LSSVM model and the mechanism model with parallel structure.The data-driven model is used to compensate the calculation deviation of the mechanism model.Simulation results show that the hybrid dynamic model has higher calculation accuracy and better adaptability of operation conditions.In addition,the widely used technology of NOx concentration detection is using CEMS system.Unfortunately,the NOx detected by this way has a large time delay,so the result cannot reflect the change of the actual NOx concentration in real time,which directly leads to the closed-loop feedback control always has a certain lag.In order to solve this problem,this paper estimates the delay time between the manipulated parameters of the boiler and the NOx concentration at SCR inlet based on Mutual Information(MI)method firstly,and then the auxiliary variables are selected through variable selection method.With the NOx concentration at SCR inlet as the target variable,a real-time dynamic soft sensor model is developed based on LSSVM to predict the NOx concentration at SCR inlet in real time.The real-time dynamic soft sensor model is used as an intelligent feedforward model,and the output of the model is converted into the feedforward signal of ammonia injection.Therefore,the intelligent feedforward control system is constructed by introducing the feedforward signal into the cascade control system to optimize the SCR ammonia injection.At the same time,in order to improve the adaptive ability of the control system,the particle optimization algorithm was used to optimize the parameters of the main controller in real time.Simulation results show that the intelligent feedforward control strategy can improve the response speed of the control system,reduce the amount of overshoot,and improve the quality of control process.Finally,based on chemical analysis and simulation software of Aspen Plus,a simple simulation of the SCR denitration reaction process is performed.First,a static model of the SCR denitration system is constructed in Aspen Plus,and the factors affecting the process of denitration reaction are simulated and analyzed.Second,the static model is converted into a dynamic model and imported into Aspen Plus Dynamics(APD).Communication module used to connect Aspen Plus with Simulink is designed as well.Finally,a control system is constructed in Simulink to control the dynamic model in APD in real time.With the co-simulation of Simulink and Aspen Plus,the real-time dynamic simulation of the SCR denitration reaction process realized,and which has important reference value for the optimization control simulation of SCR denitration system and even the system modeling and simulation of other industrial engineering with chemical reaction.