Dynamic Estimation Of NO_x Concentration At The Inlet Of Flue Gas Denitrification System For Thermal Power Units

In China’s electric power industry,thermal power generation occupies an important position,and the flue gas emission from thermal power units is one of the important factors causing air pollution.At present,most thermal power plants adopt selective catalytic reduction(SCR)technology to realize flue gas denitrification in order to reduce the emission of nitrogen oxides.At the same time,the flue gas automatic monitoring system is used to measure the concentration of NO_x in real time.However,due to its high delay in measuring the concentration of NO_x in flue gas,it can not timely reflect the change of the concentration of NO_x at the entrance of denitrification system and guide the reactor action.In this paper,SCR is used to monitor the the NO_x concentration at the entrance of denitrification reactor was studied.In order to meet the needs of different operating conditions,a dynamic prediction model was established to estimate the NO_x concentration at the entrance of R denitrification reactor.In this paper,the main factors affecting the formation of NO_x in boiler combustion system are analyzed,and appropriate variables are selected.After pretreatment of field data,feature selection measurement method based on mutual information and forward search strategy are used to select auxiliary variables with the greatest correlation with NO_x concentration and the least redundancy between variables.This method can reduce the complexity of the prediction model.On the one hand,a dynamic prediction model based on least squares support vector regression(LSSVR)with the total prediction error as the threshold and the parameters of the model modified adaptively is established to estimate the concentration of NO_x.On the other hand,based on mutual information method,the order sequence variables affecting the generation of NO_x are screened out,and the least squares support vector regression(LSSVR)is used as input to establish a dynamic prediction model,which completes the estimation of NO_x concentration at the entrance of denitrification reactor.This model makes up for the limitation that the traditional variable selection can only screen the static variables at the current time.The validity of these two methods is verified by simulation experiments.Both methods can achieve timely and accurate measurement of NO_x concentration,which is of great significance for reducing pollutant emissions of thermal power units.