Prediction of Outlet SO2 Concentration Based on Variable Selection and EMD-LSTM Network

Aiming at the problem that it is difficult to stably control the SO2 concentration at the outlet of the desulfurization system in a coal-fired power plant, a prediction model based on variable selection and empirical mode decomposition (EMD)-long short-term memory network (LSTM) was proposed. First, the relevant variables related to outlet SO2 were determined through mechanism analysis, and the LASSO algorithm was used to remove the redundant variables. Mutual information was used to determine the time delay between input variables and output variables, and time delay compensation was carried out. The compensated data was decomposed by EMD algorithm and used as the final input variable. The prediction model of SO2 concentration at the outlet was established by using LSTM. Simulation results show that Lasso algorithm removes redundant variables and improves the generalization ability of the model; EMD decomposition can extract effective information from the data and reduce the prediction error of the model; the model established by LSTM has the highest prediction accuracy and can accurately predict the change of SO2 concentration at the outlet, which is of great significance to realize the stable operation of desulfurization system.