Abstract
There are many problems when creating digital doubles. One of which is the definition of the source data: in this case, the definition of private reductions (coefficient ratios) in a continuous finishing group of stands of the broadband hot rolling mill 2000 of PJSC Magnitogorsk Iron and Steel Work. Algorithms were developed to calculate the thickness of the breakdown bar, the number of stands involved, power required for rolling and coefficient ratios to solve this problem. The algorithms are based on well-known solutions (the Imai method) using neural networks. The training of neural networks was conducted on a sample collected for the period from 01.01.2017 to 01.01.2019 work of the mill in the PyTorch library of the interpreted programming language Python. The average error εME of the calculation of coefficient ratios (according to the developed algorithms with neural networks) does not exceed 8.9% and the standard deviation σ does not exceed 0.074.
Highlights
Broadband hot rolling mills are one of the main means of production of hot rolled products
On the example of this mill, the following trends in production are observed: increased productivity; increasing requirements for the quality of rolled products, thinning of rolled products, the emergence of new high-strength steel grades, increasing the variety of assortment in one rolling companyThese trends lead to the need for innovations in production
This significantly saves time and lays the foundation for individualized mass production, since complex production routes can be quickly calculated, tested and programmed with a minimum of cost and effort. With this ‘virtual prototyping’ many problems arise, one of which is the definition of the source data for modeling physical processes. One of these problems is the determination of reduction coefficients in a continuous finishing group of broadband hot rolling mill (BHRM) 2000 stands
Summary
Broadband hot rolling mills are one of the main means of production of hot rolled products. The sources of errors or failures can be identified and eliminated even before the actual operation This significantly saves time and lays the foundation for individualized mass production, since complex production routes can be quickly calculated, tested and programmed with a minimum of cost and effort. With this ‘virtual prototyping’ many problems arise, one of which is the definition of the source data for modeling physical processes. One of these problems is the determination of reduction coefficients in a continuous finishing group of BHRM 2000 stands. For rolling production, artificial neural network models are used to predict power parameters [8, 9], thermal state of the strip and work rolls [10, 11], wear of the work rolls [12], strip profile [13], mechanical properties [14] and etc
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: IOP Conference Series: Materials Science and Engineering
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
