Neural Network Modeling And Control Of SCR Denitrification Process

With the development of society,the environment is facing severe problems and air pollution as a part of the environment problem is more and more concerned.The fact of coal-fired power plant as the main power source in our country cannot be changed in the short term.The nitrogen oxide is one of the main factors causing air pollution,so our country adopts more strict standards of emission of nitrogen oxides.It is urgent to reduce the NOx emission of the boiler effectively.SCR denitrification technology is widely used in the domestic coal-fired units,so it has a certain significance for modeling and control of SCR denitrification process.SCR denitrification process is a large delay,large inertia,nonlinear and complex process is easily affected by many factors disturbance.so the traditional method of modeling is very difficult to modeling,self-learning adaptive ability of BP neural network is highly based on applicable to nonlinear system and has certain fault tolerance.These characteristics make it is especially suitable for solving complex problems,this paper uses the BP neural network modeling method of SCR denitrification process modeling,and according to the established model,the adaptive neural network control to achieve control of the outlet flue gas concentration.Firstly,this paper introduces the research background and significance of the structure and algorithm of neural network,and modeling steps,then this paper analyzes the mechanism of SCR denitrification process,influencing factors of outlet gas concentration,determine the input and output variables,according to the field data acquisition with self in a coal-fired power plant 600 MW unit SCR denitrification system for the foundation,first selection and preprocessing of the data,then select the appropriate neural network model structure and parameters,the model is trained,then the accuracy of the collection of a part of the scene data to verify the model,results show that the error in the acceptable range,then the BP neural network model of SCR denitrification process based on the complete.BP neural network model can accurately reflect the dynamic characteristics of denitration system.Finally,the BP neural network model is established as the controlled object,using the neural network model reference adaptive control method,to realize the denitration process export NOx concentration control of SCR as the controller,using BP neural network,with the difference between output and the reference model output is reduced by modifying the weights of the neural network.The output of the controlled object reference model output can fast response,eventually making the control to meet the ideal requirements,the simulation results show that the concentration of NOx and ammonia export volume reached steady state has no obvious change,it shows that this control scheme is feasible,laid a foundation for the practical application.