In order to grasp the emission concentration of sulfur dioxide at the outlet of the desulfurization system in advance and effectively and dynamically guide the adjustment of the operating parameters of the desulfurization system facilities, the application of BP and CNN in the prediction of sulfur dioxide concentration was studied. Through the Pearson correlation analysis, the process characteristic variables with high correlation with the outlet sulfur dioxide concentration were determined. At the same time, the BP and CNN models considering the correlation analysis and the typical BP and CNN models without considering the correlation analysis are compared and analyzed in the case of 60s and 120s in advance in predicting the effect of export sulfur dioxide emission concentration. The results show that after the Pearson correlation analysis The filtered feature data is used as the input of the neural network, and the output accuracy is better improved.