抄録
In steel works, a reheating furnace is an indispensable part of a plate/hot mill process and reheats various types of slabs to its target temperature as closely as possible with tolerable uniformity. In a conventional furnace control, the only method to predict the temperature profile of a slab and to find out a proper zone temperature(index or set-point temperature) is the mathematical model. This approach has been applied to real works on condition that it has proper physical properties and logical assumption. In this study, we make a new method of the surrounding(furnace) temperature modeling to improve the accuracy of the math model for slab temperature and a improved math model which is based on this method. In general, a previous conventional math model in plate mill plant uses the simple(linear approximation) form of measured thermocouple temperature as the surrounding temperature which is appeared at the boundary condition in model. This approach causes some meaningless coefficients in the math model, in detail, the overall heat transfer coefficient. We carry out an experiment on proving this failure in the real furnace. After all, we make a new type of sub-model to make a more accurate gas temperature profile in the furnace. The math model based on this gas sub-model gives meaningful model parameters and reliable slab temperature profile, and moreover, this sub-model has the capability to show the influence of furnace conditions like a change of the temperature set-point, gas flow-rate. In detail, we tried the various forms of model for gas temperature like PCR(Principal Component Regression), A modified spline with gas flow-rate, elapsed time from set-point change. All tested forms give more reasonable result than previous one.
