NOx Emission Prediction for A Coal-fired Power Process Using Gated Recurrent Unit Approach

NOx produced by coal-fired power plants greatly affects our atmosphere environment, and makes that the NOx emission prediction for coal-fired power plants become very important. In this paper, an NOx emission prediction model for a coal-fired power process is developed using Gated Recurrent Unit approach. First, by a canonical correlation analysis method, any two variables with a higher correlation degree are grouped together and choose one of them, then four variables with the greatest influence of NOx emission are selected. Second, the model is developed by using the determined hyper-parameters. Dropout is introduced and the best value of dropout is determined by lots of experiments to prevent over-fitting. Finally, the predicted values are compared with the actual values of the power plant, and it proves that the developed model could well predict the NOx emission from a coal-fired power plant. In addition, a comparative study is carried out between the proposed model and three traditional models, our findings indicate that developed model has higher accuracy and shorter training time.