基于变量选择和XGBoost组合模型的NOx排放预测

NOx emission prediction based on variable selection and XGBoost combined model

NOx排放是燃煤锅炉的主要污染物,精准地预测NOx排放对电站锅炉燃烧优化具有重要意义。以某电厂330 MW燃煤锅炉变负荷工况数据为研究对象,提出一种基于变量选择和XGBoost组合模型的NOx排放预测方法。首先,针对复杂多样的热工变量,基于偏最小二乘(PLS)进行变量选择。然后,基于XGBoost建立NOx排放组合预测模型,组合模型通过线性模型融合基于NOx排放历史序列和变量特征的单预测模型建立。最后,通过变负荷工况数据证所提方法预测能力和泛化能力,并与其他方法对比。试验表明,所提XGBoost组合模型具有较高的预测精度和较强的泛化能力,对燃煤机组实际运行具有指导意义。

NOx emissions are the main pollutants of coal-fired boilers, it is of great significance for the optimization of power plant boiler combustion that predicts NOx emissions precisely.Taking a 300 MW coal-fired boiler in a power plant as a research object, the NOx emission prediction method based on variable selection and XGBoost combined model is proposed.First, the variable selection method is proposed to select complex and diverse thermal variables based on partial least squares(PLS).Then, the NOx emission prediction model is established based on the XGBoost combined model, and the combined model is established based on based on the NOx emission historical sequence and variable characteristics, which are combined by a linear model.Finally, the variable load condition data is used to verify the robustness of the proposed method, which is compared with other methods.The paper show that the XGBoost combined model has high prediction accuracy and strong generalization ability, and has guiding significance for the actual operation of coal-fired units.