Research On The Measurement Of Nitrogen Oxides In Thermal Power Plant

With the increasing seriousness of air pollution,the issue of NOx emissions has also received more and more attention.China’s coal-based energy structure determines the important position of thermal power,and strict restrictions have been imposed on the emission of NOx from thermal power plants,which puts higher requirements on reducing NOx emissions from thermal power plants.Therefore,the research on NOx emissions from thermal power plants is particularly important.At present,most of the model about NOx emission is static.But through mechanism analysis,there is a delay in the formation of nitrogen oxides,so static modeling is not accurate enough.Therefore,a dynamic model based on NARX neural network is proposed in this paper to model NOx emissions from thermal power plants.The superiority of dynamic modeling is verified through comparative analysis.Firstly,through the analysis of the formation mechanism of NOx,25 operating parameters affecting the formation of NOx in boilers are determined as input variables.The furnace temperature is measured inside the furnace in three layers,which improves the accuracy of temperature measurement compared with the traditional use of the tail flue gas temperature.Next,the input variables are preprocessed by factor analysis.Four common factors are obtained,replacing the original 25 input variables,which reduces the collinearity between the original input data and also reduces model calculations.Then SVM algorithm and NARX neural network algorithm are used to get the static and dynamic models of boiler NOx emission respectively.With a higher accuracy,we find that the NARX neural network have better performance compared with the static SVM model.