Artificial intelligence and expert systems in the steel industry

Abstract

This article was prepared in an effort to determine the state of the art with respect to the use of artificial intelligence and expert system technologies within the steel industry. A number of important developments have been reported and most of them resulted in significant savings. Mathematical modeling is quite important both for understanding and for controlling a process. However, most steelmaking operations are extremely complex and cannot be described mathematically. They are, however, adequately controlled by human operators on the basis of their knowledge and expertise. Because of this, artificial intelligence is an ideal technology for the automation of many steelmakingrelated processes.

An improved parameter identification schema for the dynamic model of LD converters

“…(18) and (19), applied to past castings (using the final data required in the same equations), can be considered as function of the variables (˛, ˇ) and ( ˜ ,ĩ): in detail, the following data are considered as known: C = C EP , C 0 , C IB , T = T EP , VO, VO Raff , PB, T Raff , T IB , while ˛, ˇ, ˜ ,ĩ are the variables to be determined. In particular (20) and (21).…”

“…notice that there are four unknowns (˛, ˇ, ˜ ,ĩ) and two Eqs. (18) and (19). Yet, given the dimensional homogeneity of ˛ and ˇ with ˜ andĩ, respectively, it is heuristically assumed that ˛ = ˜ and ˇ =ĩ.…”

“…Yet, given the dimensional homogeneity of ˛ and ˇ with ˜ andĩ, respectively, it is heuristically assumed that ˛ = ˜ and ˇ =ĩ. Such hypothesis allows to compute the unknown parameters as the intersection of two curves: namely the set of values ˜ andĩ satisfying (19) and the set of values of ˛ and ˇ satisfying (18) having experimentally fixed (and measured) the quantities C = C EP , C 0 , C IB , T = T EP , VO, VO Raff , PB, T Raff , T IB .…”

“…(18) and (19) can be solved considering as unknown quantities T raff and VO; in this way the quantity of coolant to be added and of oxygen to be blown are determined.…”

“…Other artificial intelligence oriented [19], rather than physical based, approaches to the modeling and control of BOF steel making process include variable selection [20] and support vector machine [21,22] techniques. Time series analysis tools for the modeling and estimation of the BOF steel making process have also been explored, for example, in [23].…”