Research On Duct Concentration Online Monitoring Method For Boiler Flue Gas Based On Reverse Modeling

Influenced by energy resources endowment, the main primary energy is coal in China. Combustion is the main way of using coal, during which, pollutants like nitrogen oxide, sulfur oxide, carbon monoxide, heavy metal and fine organic and inorganic particles will be released. Thermal power is the top priority of electricity supply in China, and pollutants control of coal-fired power plants has gained much more attention currently. With the increase investment of environment protection, on-line monitoring of dust concentration is highly desired. On-line calibration and automatic alignment of dust concentration on-line monitor will become an important supporting technology for on-line monitoring system of dust concentration. Therefore, this topic comes up with an on-line calibration method of dust concentration on-line monitor.Combined with the on-line monitoring parameters of DCS system, the selection of input variables has been realized by introducing the method of mean impact value, which helps reducing model inputting parameters, reduces the modeling complexity, improves the training velocity, and also ensures the accuracy of the model.The influence parameters of dust concentration are got based on feature selection. This paper compares and analyses the advantages of dealing with nonlinear problem with the methods of reversed modeling, such as artificial neural networks and support vector machines, and establishes an on-line monitoring model of dust emission concentration, which offers theoretical basis for the on-line monitoring of dust concentration.Finally, on-line motoring system frame is designed by taking support vector machine method as the model algorithm of on-line monitoring system, to realize the on-line monitoring and analysis on dust concentration, which lays the important foundation for the on-line calibration of on-line monitoring of dust concentration, and also the guide and reference for operators.