Technostile (Nord Clan: ML Sense)

2023: Implementation of the ML Sense system

At the same time production metalwork , the quality of the surface is important metal in order to ensure high-quality painting of future structures. For example, damage to the sheet edge is not as critical as through breaks or non-galvanizing.

To speed up the rejection process, it was necessary to detect various types of defects on the rolled metal web and automate the formation of reports.

Peculiarities of rolled metal quality control by machine vision

To determine defects on the rolled metal sheet, high-speed cameras with a resolution of 8k px by 2px are used, which determine defects with an accuracy of 1 mm. Such cameras reveal even thin scratches that make the sheet unsuitable for further painting. In case of incorrect selection of light equipment in the lighting area, penumbra and a difference in shades may occur, which the system may take as a defect. High-intensity lamps are installed for high-quality illumination of surface of rolled metal sheet. Data processing takes place using the ML Sense system developed by Nord Clan. The program is installed on a separate server, which is located in production and to protect against dust, the server is "hidden" in a protective cabinet.

Camera fixation features

Next to the rolled metal quality control post, there are risks of displacement of the cameras when installing another reel in the machine. To eliminate this danger, Nord Klan engineers specifically for the Technostil project developed a reinforced camera mount.

Smart cropping

Experts noted that when the reel moves, it shifts several centimeters to the left and right because of this, a static camera could miss defects at the edge of the rolled metal sheet. To avoid this problem, the Nord Klan specialists have developed a Smart Cropping technology that determines the edge of the sheet and frames the image according to the received one. With this, information technologies the company can also control the width of the rolled metal web along the entire length of the reel.

Defect Reports

A report is generated for each defect of a suitable size and type, which automatically displays the dimensions of the defect, its type and exact location on the rolled metal sheet. To do this, an encoder is installed on the shaft, which allows you to accurately determine at what distance the defect is located. Also, a photo of each defect obtained from a high-speed camera is automatically added to the report.

Typical defects identified by the system