基于模糊聚类和深度置信神经网络的燃煤机组NOx浓度预测方法研究OA

Aiming at the prediction of NOx concentration at the inlet of the selective catalytic reduction(SCR)reactor in coal-fired units,a NOx concentration prediction method based on fuzzy clustering and deep belief neural network is proposed.Firstly,the variables that have a greater impact on NOx concentration are selected through mechanism analysis and feature weight analysis.Secondly,the fuzzy clustering algorithm is used to partition the operating conditions of the unit,and the distribution of NOx concentration at the inlet of the SCR under different operating conditions is analyzed.Finally,based on the selected auxiliary variables,historical operating data under different operating conditions are extracted,and a NOx concentration prediction model is established based on the deep belief neural network(DBN).The method is verified by actual data from a 660 MW circulating fluidized bed boiler,and the results show that the NOx prediction model based on fuzzy clustering and deep belief neural network can accurately predict the NOx concentration in the nonlinear time-series combustion process.