Physical and chemical calculations of steelmaking processes and predictive models for the production of clean steels

The results of physicochemical calculations of steelmaking processes for the production of clean steels deoxidized with aluminum are presented. An approach is considered to improve the technology of clean steel production, including elements of mathematical and thermodynamic models, as well as algorithmic approaches for building static models using machine learning technology, which make it possible to increase efficiency in steelmaking technologies. In order to obtain an applied application of the presented approach, it is necessary to prepare a data array in advance, as well as to interpret the results of machine learning, based on the fundamental laws and physical and chemical processes occurring in steelmaking. As a result of the thermodynamic calculations performed in the STM program, measures were developed for the production of clean steels. On the examples of the production of thin slabs and billets, bloom, the search and confirmation of significant technological parameters in the formation of steel-making defects due to non-metallic inclusions was carried out using methods of in-depth analytics and machine learning.