Intelligent Detection Method Based On Multi-source Information Fusion Blast Furnace Surface Radial Temperature

The radial temperature of burden surface in Blast Furnace (BF) is an important indicator for burden surface temperature field, and a main factor to determine the distribution of radial gas flow, and it has guiding function to burden distribution of BF. However, due to the fact that BF is a high-temperature, high pressure, and sealed container, the radial temperature of burden surface in BF could not be measured directly. Aiming at these issues, detection model of radial temperature was built in this thesis, based on multi-sources information, to compute radial temperature on-line. And heat state of burden surface and distribution of radial gas flow could be reflected by this method which has an important role in steady running of BF.The characteristic and credibility of multi-sources information detection in burden surface of BF were analyzed first. Then some feature information relevant to radial temperature was determined. Because feature information could not be gotten directly, such as continuous temperature of cross temperature, radial grey curve, radial mine coke ratio. Then continuous temperature obtained by interpolation method, radial grey curve obtained by applying time-domain and spatial image filtering techniques to infrared image, radial mine coke ratio obtained from burden mathematical model. And radial temperature was computed with the three kinds of feature information.Radial temperature fusion model based on fuzzy logic was built to get radial burden temperature on this basis, and three results were calculated by weighted fusion. The weighted factors were determined by pixels gray-scale statistic method and fuzzy logic algorithm, and values of weighted factors were given in one burning cycle of BF .Finally, according to the actual run data of a 2200m~3 BF of a domestic steel company, BF detection simulation platform was established with VC 6.0, SQL2000 database and related hardware. The simulation result shows that the detection method based on multi-sources information can achieve the effective detection and visual display.