Severstal to Equip Hot Rolling Mill with Neural Network

The Directorate for Technical Development and Quality of the Severstal Russian Steel Division and Severstal Digital LLC, Metal-Expo regular exhibitors, are improving the software used at the industrial site of the Cherepovets Steel Works for quality control of the rolled metal surface.

A new in-house development of the company's specialists, the VERA neural network, is being implemented at  high-tech and most productive hot rolling mill in Russia mill 2000. VERA is an improved analogue of the EVE neural network, which was developed by Severstal employees in 2019 and is used to classify surface defects of rolled metal on the cut-to-length line # 4 in the metal finishing workshop # 2 of the flat-rolled steel production plant at CherMK.

The new program, the VERA neural network, is a similar, but more complex algorithm for finding and classifying a larger number of deviations. It will be able to find already 19 classes of defects. The program includes an anomaly detector that will allow you to identify atypical images. Also, Severstal Digital employees updated the user interface, making it more convenient and adding new functions, for example, the “like-dislike” concept, which allows the operator to point out the models to errors and correct solutions in non-standard situations. This approach allows you to adapt to changing conditions: the model is periodically retrained on the basis of operator corrections and adapts to the technological process.