火电厂NO_(x)的均布优化及预测研究

RESEARCH ON UNIFORM DISTRIBUTION OPTIMIZATION AND PREDICTION OF NO_(x) IN THERMAL POWER PLANTS

某330 MW机组脱硝改造前,A、B侧脱硝反应器的出口NO_(x)浓度分布均匀性均较差,脱硝改造后,100%ECR工况条件下流线在整个SCR脱硝系统中的分布比较均匀,第1层催化剂入口5个分区内的NO_(x)浓度分布相对标准偏差分别为1.73%、1.03%、0.35%、0.36%和1.17%。采用LSTM模型、SVR模型、LSTM-SVR混合模型,将火电厂DCS数据与时间序列相关联,对NO_(x)浓度进行预测。结果表明:LSTM-SVR混合模型与LSTM模型对脱硝入口及出口的NO_(x)浓度的预测效果均较好。现场验证测试发现脱硝改造后脱硝反应器入口、出口的NO_(x)浓度分布比较均匀,用于预测分析用的DCS数据与实测数据基本吻合。

Before the denitration transformation of a 330 MW unit,the uniformity of the NO_(x)concentration distribution at the outlet of the denitration reactors on the A and B sides was poor.The relative standard deviations of NO_(x)concentration distribution in the five zones at the inlet of the layered catalyst are 1.73%,1.03%,0.35%,0.36%and 1.17%,respectively.The LSTM model,the SVR model,and the LSTM-SVR hybrid model are used to correlate the DCS data of the thermal power plant with the time series to predict the NO_(x)concentration.The results show that both the LSTM-SVR hybrid model and the LSTM model can predict the NO_(x)concentration at the inlet and outlet of the denitrification well.The field verification test found that the NO_(x)concentration distribution at the inlet and outlet of the denitrification reactor after the denitration transformation was relatively uniform,and the DCS data used for prediction and analysis were basically consistent with the measured data.