Oskol Electric Steel Works completed tests of an automated system of recognition of a brand on cast procurements
Customers: Oskol Electric Steel Works (OEMK) Stary Oskol; Metallurgical industry Contractors: NITU MISIS (National University of Science and Technology), metallo-Tekh (before JSA Group, Dzhi Es Ey Group) Product: Video analytics (projects)На базе: Complex projects of video surveillance Project date: 2020/02 - 2020/08
|
2020: Completion of tests of an automated system of recognition of a brand on cast procurements
Oskol Electric Steel Works (is a part of Metalloinvest company) completes trial operation of an automated system of recognition of a brand on cast procurements in sortoprokatny workshop No. 1. Ickx Holding reported about it on September 25, 2020. The idea of the project came up at the staff of OEMK during the competition of innovations organized by JSA Group (enters into the multi-profile IT group "ICKX Holding") at the enterprise.
Own intelligent solution on the basis of neural network technologies is developed by scientific STI NITU "MISIS" jointly with specialists of plant. Software engineers of Ltd company "metal - That" were responsible for integration of a program part of a system into production processes.
Earlier the operator manually verified brand digits with the list in a system, selected the furnace for melting and then sent to it procurement, – the supervising foreman of SPTs No. 1 Evgeny Tkach tells. – The employee had no more than 40 seconds on accomplishment of these tasks that created a heavy load and raised an error probability. Implementation of automatic recognition significantly reduced these risks. |
The implemented solution uses five neural networks, each of which solves the separate problem. In the course of data collection for their training more than 60 thousand pictures of end faces of procurements with a brand were made.
One option of a brand on procurement is printed by the machine, another, in case of correction of number, – people manually paint, – Dmitry Poleshchenko, the associate professor automated and information management systems of STI NITU "MISIS" explains. – At first sight process of recognition is similar to reading of numbers on cars. But in the analysis the similar systems did not yield required result. |
This system became the first product developed within the program of digital transformation of the company. This solution together with colleagues from MISIS was created by our internal command, – the director of digital transformation of Metalloinvest Management Company, the vice president of "IKS of Holding" Yulia Shutkina noted. – We will continue headed for development of the Russian technologies in production. |
The experience got in implementation process of this project is going to be applied to the solution of similar tasks. For example, the technology of video analytics can be implemented in determination of granularity of structure on the pipeline, for determination of quality of tooth of a bucket of the excavator and even for tracking of use by the staff of the enterprise of individual protection equipment.