The Boiler Flame Temperature Measuring Research Based On Digital Image Processing

As to large-scale coal-fired boiler of power station, controlling the flame temperature in the stove are very important for the dependability, security and economy operated to the boiler. The combustion process of the coal powder boiler is that a kind of very complicated heat radiation exchanges course, it's operating mode is extremely unstable. The security that the boiler burns depends on the stability of flame mainly, in order to avoid the emergence of the explosive incident of the boiler, the coal powder must be burnt steadily in the stove, if it is unstable to burn, temperature field is not even in the stove, light to cause thermal efficiency of boiler reduce greatly, cause to waste, heavy and easy to is it produce stove serious consequence to present. So, the accurate measurement and forecasting of flame temperature field in the stove is very important for the burning state judgement, and adjust the operation parameter of the boiler in view of the above, it is very important to ensure the normal running of the boiler.This text has recommended the working mechanism of the coal-burning boiler and existing temperature predict principle, further investigated the temperature survey method based on colored CCD flame picture. In order to improve the accuracy based on colored CCD flame temperature field measurement, on the basis of analysing error source, compare multinomial interpolation, least square method and BP neural network application in reducing the error of temperature survey. Direct against the flame picture of stove, having analysed the source of the noise of the picture, a large number of random noises in following to the combustion process have been successful dealt with in using median filtering, have realized the gray transformation and false color shown of the flame picture. In order to predict the variation tendency of fire box temperature accurately, improve production efficiency, this text has proposed using the forecast model of flame temperature in the stove based on BP neural network. discussed the choice of the forecast model's input parameter and date pretreatment technology.The full text is divided into five chapters, chapter one recommends the commonly used high-temperature flame measure methods at present, including contact-type temperature measure method and non-contact-type temperature measure method, explain CCD flame temperature detection method based on optical radiation energy. Chapter two explain the structure the flame temperature measuring system...