Advanced diagnostics of industrial pulverized coal burner using optical methods and artificial intelligence

Abstract

Controlling the combustion process is a very complex issue. The difficulty of operation of such process consists in the mutual interference effects of chemical, physical (mainly energy) and mechanical nature on one hand and risks existing if its course becomes unpredictable. In addition, there are restrictions on the control due to the unavailability of certain process signals (input or output) and incomplete knowledge about them. Current availability of high-speed measuring and computing devices allows to extract the hidden relationships between the elements of such complex process and the use them in control. The paper presents the technologies being developed in the Department of Electronics Lublin University of Technology. They use optical diagnostic methods and artificial intelligence methods.Research is aimed to develop a system allowing a parametric evaluation of the quality of pulverized coal burner operation. It is based on an analysis of local variability of the brightness of the flame. Due to the highly nonlinear nature of dependency and lack of an analytical model, fuzzy-neural methods were used to estimate the selected parameter.The studies, described in the article, confirm that in order to obtain NOx emissions from pulverized coal burner the estimate calculated on the basis of immediate optical signals can be used instead of the delayed signals from the gas analyzers. The use of neuro-fuzzy models allows to determine emissions of nitrogen oxides with satisfactory accuracy and time, what allows application in control systems.