Prediction of Blast Furnace Temperature Based on Multi-information Fusion of Image and Data

摘要

Tuyere CCD images reflect the thermal state of Blast Furnace (BF) hearth and represent its temperature change. It is the most direct and timely information when blast furnace operators judge the furnace temperature. However, early furnace temperature prediction models have not considered the tuyere images, whose prediction precision is lower. This paper collects a steel mill's 2500 m^{3 blast furnace online data and its tuyere images, establishes time series neural network models based on multi-information fusion, which compares with other three different models only based on BF data. The simulation results prove that tuyere images can effectively improve the realtime and the precision of the hearth temperature prediction model.}