| 摘要 |
PURPOSE:To effectively predict the temperature drop of molten steel in a ladle by learning in advance a correction value from a difference between an actual result value and a predictor in every ladle, and correcting the predictor by its correction value at the time of using the same ladle in the next time. CONSTITUTION:Surface temperature of a refractory and a shell is measured by using a radiation thermometer, etc., and with regard to plural charges, its temperature is measured repeatedly and recorded. As a result, based on a relation of a surface temperature of an inner wall of a vessel refractory determined based on collected data, thickness of the vessel refractory, and whether covering is executed or not after the vessel becomes empty after pouring is finished, an internal temperature distribution of the vessel refractory is derived from the measured inner wall surface temperature of the vessel refractory. Subsequently, by deriving an average temperature of the vessel refractory, based on this temperature distribution, a heat accumulation amount of the refractory and a heat removing amount by the vessel which is being carried are estimated. Accordingly, by learning the heat accumulation amount of every vessel, and executing a correction by using result of learning at the time of deciding the heat accumulation amount of the vessel in the next time, the prediction of high accuracy can be executed. |
