The design of boiler combustion diagnostic methods and real-time system

Due to the boiler burning supervisory system´s higher demand of the real-time and stability, this paper introduces the design of a flame image processing system based on PCI bus and DSP. The discussion reflects the characteristic parameters of the flame image and the burning diagnosis method based on neural network. It detailed analysis and explains parts circuit and workflow about the image acquisition and processing, and adopts an optimization design of the combination of PC computer and TMS320C6205, which greatly improves the system´s data processing ability, real-time and accuracy of ascension. After 200MW unit test shows that, in one frame, system can finish an image processing and characteristic extraction, and in 5 seconds, system can complete the diagnosis of boiler combustion status, which ensures the real-time character of flame monitoring and burning diagnosis.