Based On Parameter Estimation Fusion Blast Furnace Surface Temperature Field Modeling

With the problems of complex internal environment of blast furnace, elusive physical, chemical, dynamic process, diverse temperature distribution of burden surface, and hard-established model of temperature field of burden surface, an on-line detection way of temperature field of blast furnace model is clarified in this thesis. Based on analysis of the relationship of Infrared image and cross- temperature from the technologic point of view, temperature field model which is based on infrared images and the cross-temperature is brought forward.Firstly, based on analysis of characteristics of infrared camera of blast furnace, a temperature field of blast-furnace model which is based on infrared image is established by means of wavelet transform algorithm image segmentation and processing of high-frequency and low frequency components of image. Meanwhile, based on analysis of the characteristics of cross-temperature, a temperature field model of cross-temperature is established by way of using weighted average algorithm.Secondly, the temporal registration of two models, which using least square method, and detection units of two model that using dynamic calibration algorithm is made aiming at the problems of inconsistent of infrared image based model and cross-temperature model in terms of time and measurement unit. On the basis of this, the method of parameter estimation based high-expectation and iterative calculation through the E-step and M-step is used to generate the final on-line detection temperature field model of blast furnace.Finally, Monitoring system of temperature field of blast furnace is established by using modeling method above. Visual interface is adopted to show the temperature field of burden surface. Further more, it can reflect the distribution of surface temperature for the blast furnace and provide blast furnace operators with real time and reliable reference information.