Pokročilé metody analýzy dat v řízení průmyslových systémů

Abstract

The topic of the dissertation is focused on the use of advanced methods of data analysis with emphasis on the use of these methods and procedures in the control of industrial systems with especially in the metallurgical industry. The concept of this work is based on data mining, which can be perceived as a set of tools, procedures and methods that can be used to obtain information and possibly knowledge from data, which are very often collected by individual technological units in industry, but are no longer used. The research mentioned below clearly indicates that these data hide a wide range of informations that can be used to optimize the operation of these technologies, or to create models for control of these technologies. The main goal of the dissertation is to design a methodology that will be applicable in the metallurgical industry for the processing of operational data, which so far remain useless. It is based on the methods used so far, which are analogously adapted to the specifics of the metallurgical industry, especially the large volumes of data that are used here. Innovative in this area is the preparation of a methodology for the use of current data processing tools, namely the use of neural networks, which should allow quality data analysis, draw adequate conclusions, even in cases where standard identification methods or fuzzy systems can not be used. The main goal of the work was then verified by applying the methodology to a specific operation in the field of metallurgy. This work contains the main results of data analysis and their interpretation.