SCR烟气脱硝系统自适应混合动态模型

随着火电厂环保要求的不断提高,选择性催化还原(SCR)烟气脱硝系统得到了广泛的应用。准确的动态模型是SCR运行优化的基础,在机理模型的基础上,利用核偏最小二乘(KPLS)模型修正机理模型偏差,构建了SCR脱硝系统的自适应混合动态模型。利用现场实际运行数据验证了模型的有效性,并与机理和数据模型进行了对比。对比结果表明,自适应混合模型结合了机理和数据模型的优点,准确描述了SCR脱硝反应的动态过程,具有更高的拟合和预测精度。

Self adaptive hybrid dynamic model of SCR flue gas denitration system

With the continuous improvement of environmental protection requirements, selective catalytic reduction (SCR) flue gas denitration systems have been widely used in thermal power plants. The accurate dynamic model is the basis of the optimization of SCR operation. In this paper, a self adaptive hybrid dynamic model of SCR denitration system was constructed based on the mechanism model. In the proposed model, kernel partial least squares (KPLS) model was used to correct the deviation of the mechanism model,. The validity of the model was verified by the actual operation data, and the model was compared with the mechanism and the data driven models. Comparison results show that the hybrid model combines the advantages of mechanism and data driven models, and describes the dynamic processes of SCR denitration reaction accurately and has higher fitting and prediction accuracy.