Data-driven thermal efficiency modeling and optimization for reheating furnace based on statistics analysis

The rolling reheating furnace is widely used in the large-scale iron and steel plant and the constantly changing dynamic characteristics and the interaction effect between different heating zones become challenges to operate a reheating furnace in an efficient way. In this paper, statistics analysis methods are utilized to justify the significance of the derived variables for the thermal efficiency modeling. By employing nonnegative garrote (NNG) variable selection procedure, an adaptive scheme for thermal efficiency modeling and adjustment is proposed and virtually implemented for a rolling reheating furnace. The detail analysis results show that there is good control precision improvement and large energy-saving benefit when the furnace operation shifts from the present practice to the model-based optimization adjustment.