基于DTW-CNN-LSTM的SCR出口NO_x浓度预测

Prediction of NO_x Concentration at SCR Outlet Based on DTW-CNN-LSTM

摘要: 为了解决火力发电厂SCR烟气脱硝设备操作人员依赖经验调节喷氨阀门开度，降低SCR脱硝系统出口烟气中NO_x浓度的问题，提出了一种SCR出口NO_x浓度预测方法。该方法基于一维卷积神经网络和长短时记忆神经网络，同时在数据处理过程中加入动态时间规整算法，利用青岛某热电厂发电机组的SCR脱硝系统运行数据，搭建SCR脱硝系统出口NO_x浓度预测模型，通过提取数据在时序上的特征，可实现预测45分钟内SCR出口NO_x浓度。电厂SCR脱硝设备操作人员可将该模型的预测结果作为调节阀门开度时的重要参考，将其调整至当前最佳状态。结果表明，DTW-CNN-LSTM模型在SCR出口浓度预测精度优于传统LSTM及CNN-LSTM，在测试集上$R^2$ 为79.65%，得到了期望的结果。

Abstract: In order to solve the problem of operators of SCR flue gas denitrification equipment in thermal power plants relying on experience to adjust the opening of the ammonia injection valve and reduce the concentration of NO_x in the outlet flue gas of the SCR denitrification system, a method for predicting the NO_x concentration at the SCR outlet is proposed. This method, based on a one-dimensional convolutional neural network and long short-term memory neural network, incorporates a dynamic time warping algorithm in the data processing process. Using the operating data of the SCR denitrification system of a power plant in Qingdao, a prediction model for the NO_x concentration at the outlet of the SCR denitrification system is built. By extracting the temporal features of the data, the NO_x concentration at the SCR outlet can be predicted within 45 minutes. The operators of SCR denitrification equipment in power plants can use the predicted results of this model as an important reference when adjusting valve opening, and adjust it to the current optimal state. The results showed that the DTW-CNN-LSTM model had better accuracy in predicting SCR outlet concentration than traditional LSTM and CNN-LSTM, with R-Square of 79.65% on the test set, achieving the expected results.