Sidanchenko V. Automated method for estimating and predicting of the chemical composition of cast iron at the output of the blast furnace

The dissertation is a completed scientific research work, which is devoted to the development and justification of new methods of automated estimating and forecasting of the chemical composition of cast iron at the output of a blast furnace. In this dissertation, on the basis of the obtained theoretical and experimental results, the actual scientific and applied problem of creating and researching methods of automated control and forecasting of the chemical composition of cast iron at the output of the blast furnace was solved, which made it possible to increase the accuracy and reliability of the forecast of data on the chemical composition of cast iron, to increase efficiency management of blast furnace smelting and, therefore, to ensure the specified product quality while reducing material and financial costs for its production. The aim of the dissertation is to improve the quality of control of the iron smelting process by developing new estimating and forecasting methods and tools, adequate to the nature of the ongoing processes, which allow increasing the accuracy and reliability of estimating and forecasting of the chemical composition of cast iron at the output of the blast furnace. The object of research is the process of smelting iron in a blast furnace. The subject of the research is the methods, means and models of creating automated control systems and forecasting of the chemical composition of cast iron at the output of the blast furnace. The thesis consists of an introduction, three chapters, general conclusions, a list of used literary sources with 99 items and 4 appendices. In the introduction, the relevance of the research topic is substantiated and formulated the purpose, object, subject and main tasks of the research, the research methods are justified, the scientific novelty and practical significance of the obtained results are explained, the personal contribution of the recipient is indicated, the general characteristics and structure of the dissertation are presented, as well as information about publications and the results of the approval and implementation of the dissertation work . In the first chapter of the dissertation, an information analysis of the current state of the problem of automated control and forecasting of the chemical composition of cast iron is performed. "Classical" methods and models of processing stochastic processes with Gaussian distribution are considered. Methods and models of nonlinear dynamics and deterministic chaos are considered. It is hypothesized that the existing methods of estimating and forecasting the chemical composition of cast iron, based on the Gaussian nature of stochastic time series that provide data on the chemical composition, are inadequate to the real law of distribution. Formulated formulation of the research problem. In the second chapter, a study of the stochastic properties of data (time series) on the chemical composition of cast iron is carried out. The non-Gaussianity of the time series data was proved and the hypothesis about their fractal character was substantiated. Further studies of the fractal properties of time series unequivocally confirmed the proposed hypothesis. It was established that the autocorrelation function has a so-called "heavy tail", that is, it does not tend to zero when the observation time tends to infinity. The Hurst index calculated using RS-analysis was of the order of 0.6189, which indicates the self-similar property of the fractal process. The analysis of time series by the methods of nonlinear dynamics and deterministic chaos made it possible to obtain fundamentally new information (the analysis of literary sources did not allow the author to obtain information about similar studies due to their absence) about the properties and characteristics of the investigated time series.