Chapter 9 - Knowledge Based Modeling

Abstract

Listen

Translate article

Knowledge based modeling, including the use of artificial intelligence, expert systems, fuzzy sets, and logic and neural networks, is an exciting approach to analyze the rolling process. This chapter presents the possibilities, starting with several definitions of terms, not commonly used in engineering. The importance of acquisition of knowledge and its storage are outlined. Data mining and the use of self organizing maps in the hot rolling process are also demonstrated. The chapter also presents case studies, in which the use of the techniques for the prediction of the quality of the product using the methods of artificial intelligence and the use of self-organizing maps in a hot strip rolling mill is discussed and applied to several problems associated with the rolling process. A frame work, to allow the use of neural networks in the prediction of the grain size in hot rolling, is the first example in the case study. This is followed by the application of neural networks to the prediction of constitutive behavior of steels and aluminum and to the prediction of the roll separating force during hot rolling of strips of the two alloys. The chapter then highlights that the major concern in the knowledge based process is the closed logical loop in the technological design, because every technological operation can influence the quality of the product. Technology affects the material properties that further influence the geometry that determines the design of the technology. Disciplines of mechanics and kinetics cannot easily handle the problems of friction, wear, fracture, metallurgical phenomena, and heat loss and gain in multistage processes. Modeling such sophisticated phenomena needs new approaches. For process planning, measured and predicted data are also necessary.