Research of flame image processing and control system

This article describes an furnace flame closed-loop control system based on image signal processing. Through the power plant furnace flame image processing, obtained the rate of effective temperature area, the rate of high-temperature area, centroid offset distance. Taking the data as criterion, we used KOHENON neural network to distinguish the state of flame and fused the related signals of sensors such as flame detector. Then proposed the incremental opening adjustment in order to complete real-time control of the furnace flame.