Optimization Of SNCR Denitration Control Process In Circulating Fluidized Bed Boiler

Nitrogen oxides released by the combustion of coal in thermal power plants are the main components of air pollutants.In recent years,the state has continuously lowered the nitrogen oxide emission limits of coal-fired power plants,which has increased the cost of denitrification for power generation companies.In the practical application of selective non-catalytic reduction technology in circulating fluidized bed power plant,there are problems such as long delay time of online measurement of nitrogen oxides and poor control effect of denitrification system based on PID control.It is of great significance for enterprises and countries to realize accurate and efficient control of denitration system and reduce denitration cost.Aiming at the problem of long delay time in online measurement of nitrogen oxides,a nitrogen oxides concentration prediction model based on Long Short-term Memory network is established.Firstly,the delay time of nitrogen oxides concentration measured by CEMS system is obtained by analyzing the nitrogen oxides data and purge signals in historical data.Then the m RMR algorithm is used to analyze the correlation of the selected variables after data preprocessing,and seven groups of variables with strong correlation with nitrogen oxides are selected as the input variables of the prediction model.Furthermore,the super-parameters of the Long Short-term Memory network model are determined by particle swarm optimization algorithm.Finally,the nitrogen oxide concentration prediction model of circulating fluidized bed flue gas is established by using the data of corrected delay time.According to the comparative analysis of the prediction results,the root mean square error of this method is reduced by 32.05 %compared with the measured value,which improves the problem of large measurement error caused by the measurement delay.To overcome the problem of poor control effect of PID control denitration system,the multimodel control system based on stair-like generalized predictive control is used to optimize the control effect.Firstly,the controlled object is divided into four different sub-models according to the load.Then,the transfer function of each sub-model is identified offline,and the corresponding stair-like generalized predictive controllers are designed respectively.The multi-model control based on stair-like generalized predictive control is used to optimize the original denitrification control system.The simulation results of the four sub-models show that the control effect of the stepped generalized predictive controller is more stable than that of the generalized predictive control.Compared with PID control,the Settling time of stair-like generalized predictive control can be reduced by 47.44 %,56.93 %,56.58 % and 55.31 % respectively,which effectively improves the reaction speed of the control system and optimizes the control effect.Based on the original SNCR denitration control system,a SNCR denitration optimization control system for circulating fluidized bed boiler is designed.The system builds the hardware platform of the denitration optimization control system.Based on the hardware platform,the software platform of the denitration optimization control system is developed by using MATLAB with the human-computer interaction layer,functional logic layer and data communication layer as the main architecture.On the basis of keeping the structure of the original denitration control system unchanged,the proposed denitration optimization control system uses the stepped generalized predictive control to optimize the control process,which provides a new method for the optimization and transformation of the SNCR denitration system of the circulating fluidized bed boiler.