Application of improved BP network in flame burning condition recognition

Whether the powdered coal boiler chamber burning flame is stable or not is an important condition for the boiler security and the economical movement. Therefore, the prompt reliable flame examination technology is the basic demand for the power plant safe operation. According to the flame images gathering from the tangential burner and the swirl burner, the gradation mean value and standard of two characteristics vectors in the flame picture characteristic area are refined based on the digital image processing technology and the discussion of withdraws method and the significance of the characteristic value. Applying modern artificial nerve network intelligence theory, the BP network algorithm is designed and improved. After the process training and the practical application, the BP network has shown the good recognition capability to the certain operating mode eddy burner and the direct current burner flame burning condition. Moreover, the distinction is extremely accurate, and the network is stable, which has obtained the good practical application effect in the scene.