Прогнозирование содержания углерода в полупродукте на выпуске из большегрузных 350-т конвертеров с применением непараметрической статистики

The presented material presents the results of using non-parametric statistics methods for predicting the carbon content in a steel-casting ladle made of two simultaneously working 350-t converters with a carbon semi-product. The situation is analyzed when the variables – the temperature of the melt...

Full description

Saved in:
Bibliographic Details
Date:2023
Main Authors: Бондарь, В. И., Мельник, С. Г.
Format: Article
Language:Ukrainian
Published: Physico-technological Institute of Metals and Alloys 2023
Subjects:
Online Access:https://www.metalsandcasting.com/index.php/mcu/article/view/103
Tags: Add Tag
No Tags, Be the first to tag this record!
Journal Title:Metal and Casting of Ukraine

Institution

Metal and Casting of Ukraine
Description
Summary:The presented material presents the results of using non-parametric statistics methods for predicting the carbon content in a steel-casting ladle made of two simultaneously working 350-t converters with a carbon semi-product. The situation is analyzed when the variables – the temperature of the melt, data on its chemical composition, as well as the degree of its oxidation, are connected by stochastic dependence. The features of power-law models for the determination of carbon during the conversion of a metal bath are analyzed. The results are obtained using non-linear regression analysis methods in the Fixed non linear mode in the form of non-linear regression equations. The adequacy of the best regression equations is confirmed. They account for at least 80 % of the variation of the independent variable. Regression coefficients are significant at a significance level of not more than 0.05. The standard error in estimating the dependent variable – the carbon content in the metal melt, when using the obtained regression equations is no more than 5 % of the average value of the dependent variable. Residues from the regression are without noticeable systematic autocorrelation, are normally distributed and independent of each other. The possibility of using the presented model for predicting the carbon content in a carbon intermediate was established, provided that two 350-t converters working simultaneously, with tuyere-probes, were needed without the need for laboratory chemical analysis of melts. This should contribute to the production of steel of a particular grade with a given chemical composition and temperature. Based on the results obtained, recommendations were developed and proposed for their practical use in the production of converter steel.