Alabuga-Fiber (Nord Clan: ML Sense)

2023: Nord Clan's machine vision and neural network-based fiber quality control system ML Sense

LLC Alabuga-Fiber has introduced a system for controlling defects in carbon fiber based on machine vision and neural networks (ML Sense from Nord Clan).

When producing carbon fiber, about 400 bundles move along the conveyor at the same time, which complicates visual quality control by QCD employees. The finished ML Sense system from Nord Clan is able to simultaneously track such a number of threads on several shafts up to 3 m wide at a speed of up to 12 m/min and at the same time give a defect detection accuracy of at least 97%.

Representatives of the Nord Clan implemented the ML Sense system based on the peculiarities of the enterprise and took into account the production cycle of carbon fiber production. The commissioning of the quality control system took place on a turnkey basis: selection of drawings and layouts of the equipment and installation, integration of the finished system with the equipment of Alabuga-Volod LLC, commissioning and training of users to work with the system. The implementation took place without stopping the company's technological processes.

High-speed machine vision cameras and a special lighting system are installed on the fiber production line. The camera transmits a video stream to the ML Sense system, which in turn recognizes typical fiber defects: pile, break, knot, thickness change, foreign inclusions (garbage, insects, resin, etc.).

Information on the detected defects is transmitted to the AWS of the QCD operator - a special information kiosk, which displays data on the detected defect. DETAILED information about the defect is displayed on the screen:

• name of the defect;
• defect class;
• sequence numbers of bundles;
• date and time of defect fixation;
• image of the defect.

In addition, a light and sound alert is installed on the production line. The system signals the operator about defects in the carbon fiber production line. The operator can detect the defect in a timely manner and localize it: determine the place of arrival of the defective bundle on the receiving coils and replace or mark the defective coil, eliminate the defect.

As a result of the implementation of the system, Alabuga-Fiber LLC launched a single marriage control center, which is integrated with the MES system. The implemented ML Sense solution in real time collects information about all defects, saves the history and generates reports, provides automatic unloading of information to the MES system of the enterprise.

This allowed us to increase the yield of finished products that meet quality standards, reduce scrap costs and the number of complaints from customers, as well as reduce the company's reputation risks.