基于SVR-GA的燃煤锅炉NOx含量软测量研究

燃煤锅炉脱硝反应器入口NOx浓度是锅炉燃烧过程中的一个重要参数，针对常规NOx分析测量仪存在的精确度较低的问题，通过支持向量回归(SVR)算法建立锅炉的NOx排放预测模型，选取均方误差(MSE)作为模型的评估函数，利用遗传算法(GA)对模型参数进行优化，编制程序对NOx排放量进行预测。以某300MW燃煤机组现场数据为基础进行仿真验证，结果表明，在与SVR、BPNN等方法的对比中，GA-SVR所建立的模型在NOx排放量的预测中取得了更优的效果。

A Study on the Soft Sensorfor NOx Content of Coal-fired Boiler Based on SVR-GA

For the NOx concentration at the inlet of denitration reactor for coal-fired boiler, an important parameter in the combustion process for the boiler, the precision of conventional NOx analyzer is low. Therefore, a NOx emission prediction model for the boiler was established based on the support vector regression (SVR) algorithm, with mean square error (MSE) as the evaluation function for the model. The genetic algorithm (GA) was adopted to optimize relevant parameters of the model, and the NOx emission was predicted through programming. A simulation verification was carried out based on the field data of a 300MW coal-fired unit, and the results showed that the model established by GA-SVR was more effective than SVR and BPNN in the NOx emission prediction.