A novel method based on information fusion for modeling burden surface temperature field in blast furnace

A novel method for modeling burden surface temperature field in blast furnace (BF) is presented in this paper. It is based on information fusion and makes full use of temperature information detected from the throat of BF. A dynamic temperature calibration method based on two-point method is taken as the benchmark calibration method, in which the nonlinear error involved is adjusted by an improved BP neural network based on genetic algorithm (GA). Due to the high robustness and better nonlinear approximation ability, the proposed method overcomes the shortages of conventional calibration methods, and gets the distribution model of temperature field with more details. The simulation results and industrial implementation show that, this model of burden surface temperature field can depict distribution of temperature field more exactly, so that it is more effective to understand the distribution of gas flow and instruct the operation of burden distribution.