Yevgen Nasadyuk

Abstract

Hot rolling mills - blooming and slab mills intended for the production of billets of steel ingots, - historically deemed not suitable for automation facilities [1]. However, the realities of the economic situation are such that, nowadays most of the locally produced products output in the iron and steel industry started in blooming and slabing mills. Along with the reconstruction of the mechanical parts and electrical equipment, it became a pressing need to automate. This requires the development of mathematical models and quantitative description of industrial processes [2].

Figure 1. Animated hot rolling mill

This is developed by leading scientific production associations and researched institutes. Among domestic companies best result has SPA "Donix", that introduce automation into metallurgical enterprises of Ukraine [4]. On past soviet area the greatest experience belongs to AKH VNIIMETMASH [5]. But all this companies do not use the methods of artificial intelligence in automated systems. Unlike the world's leader of automation SIEMENS AG, that in recent products use combined system for the calculation and prediction of the properties and characteristics in all stages of production of metal [6].

One common case that is being popular is the application of neural networks for the calculation of correction factors to be used in traditional mathematical models that control a given process. That is, a hybrid traditional mathematical process-neural networks, where the neural network act as an assistant to the mathematical model.

Figure 2. Diagram of working

This distrust on neural networks arises from many factors. First of all, only recently this technique became feasible, with the increasingly wide availability of low-cost computer power. Besides that, as the mathematical foundations of neural networks are not still completely developed, no one knows exactly the mechanisms of its learning, that is, it is unknown how a neural networks calculates a given result. So, frequently they are considered as potentially unreliable "black boxes". Under some aspects this is a justified attitude, as a wrong decision taken by an industrial automation system can lead to disastrous failures and high economical losses.

References
1. Babayev F.V. The optimum cutting material using computers. - M : Machinery, 1982.
2. Technology instruction 234-P.03.01 -2002. Production of blooms, billets and slabs of carbon and alloyed steels in blooming mill department. JSC “Donetsk Metallurgical Plant (DMZ)"
3. Portal of Masters of DonNTU http://masters.donntu.ru/2006/kita/papko
4. Web-site SPA "Donix" www.donix-ua.com
5. Official site AKH "VNIIMETMASH" www.vniimetmash.ru
6. Site Siemens AG www.siemens.com