Improved BP network-based real-time monitor for NO/sub X/ releasing from power plant furnace

NO/sub X/ is a primary component of contamination gas releasing from coal, it must be monitored availably. Because the burning process of coal powder in furnace is a kind of very complex suspension combustion, the monitoring system image includes a lot of noise signals from different hands. To eliminate these noise signals of image, a kind of new median filter is introduced. The improved BP network structure is discussed in detail, in which the eigenvectors are got from flame image and the parameters of sample data are got from DCS, the initiating connection strength and the algorithm improvement are based upon nonlinear and non-restrict theories. It has been successfully applied to monitoring the total amount of NO/sub X/ emission.