Prediction of NOx emission concentration from coal-fired power plant based on joint knowledge and data driven

Accurate NOx concentration prediction is of great significance for the pollutant emission control and safe operation of coal-fired power plants. The global properties of the research object cannot be adequately described by a single data driven model, which hinders generalization performance. We propose a NOx emission concentration prediction method based on joint knowledge and data driven. First, we introduce a knowledge driven combined feature selection method to provide a global feature basis for data driven modeling. Second, we enable adaptive decomposition of the variational modal decomposition (VMD) using the modal energy difference and sample entropy. The method can extract deep time-frequency information in nonlinear and non-smooth features. Finally, we use the Informer combined with an adaptive time series segmentation method to predict NOx concentration. The experimental results indicate that the proposed method predicts the NOx concentration better than several comparative models.