Dynamic modeling for SO2-NOx emission concentration of circulating fluidized bed units based on quantum genetic algorithm - Extreme learning machine

Abstract

To meet the stringent emission standards and achieve cleaner production of circulating fluidized bed units, it is necessary to build a dynamic model of pollutants emission for creating an economical and environmentally friendly pollutant removal operation mode. This article fources on the modeling and accurate prediction of SO2-NOx emission concentration of the circulating fluidized bed unit. According to the generation and reduction mechanism of pollutants, the model inputs are selected and determined by Pearson coefficient. Then, a dynamic model of SO2-NOx emission concentration based on extreme learning machine is developed, and quantum genetic algorithm is used to optimize the connection weight between the input layer and the hidden layer and threshold of the extreme learning machine, which contributes to increase prediction performance. The test result shows that the optimized model can effectively imitate the dynamic trends of actual measured data with appropriate accuracy, the mean absolute percentage error of SO2 concentration and NOx concentration are 4.63% and 3.09% respectively. And the model's satisfactory generalization ability is demonstrated based on the generalization experiment. In addition, compared with other methods, the proposed modeling method for SO2-NOx concentration has more suitability and accuracy under dynamic conditions, which contributes to online optimization of pollutant control and intelligent development of circulating fluidized bed units.