RUSAL has announced the deployment of its neural network to monitor the quality of the aluminium ingots produced at the Irkutsk Aluminium Smelter.

The new neural network system automatically checks the surface quality of small aluminium ingots during production, before the ingots are packed for shipment to customers, assuring the quality of the products produced at the smelter. With the technology, any surface defects on any individual aluminium ingots can be detected on the casting conveyor - using a video monitoring system, that will be integrated into the existing monitoring system, and special-purpose software.

The system detects various types of defects, such as cracks, bumps, and foreign inclusions. Whenever a defect is found, the ingot is labelled as defective by a laser, and is then removed from the conveyor. All the information concerning the defects is saved and displayed as an analytical diagram to log each day’s data, regarding the number of each type of defect per day and the number of ingots that did not pass quality assurance per shift. The data is then used by employees to tweak the parameters based on which ingots are found to be defective improving the product.

RUSAL’s Chief Technical Officer, Victor Mann, commented on this new software,

“This new solution enables production line personnel to track the quality of aluminium ingots in real time, while they're being cast. It will guarantee that the products which get shipped to our customers meet all the applicable requirements”,

RUSAL’s Chief Technical Officer, Victor Mann

The Company plans for its other aluminium smelters to also integrate the new neural network to further improve the quality of its products. Thus, the deployment of robotics systems at several of the Company's aluminium smelters has completely automated the task of stacking finished aluminium ingots onto pallets, moving the company into the future of Industry 4.0.

For more information, visit:

https://rusal.ru/en/press-center/press-releases/rusal-deploys-neural-network-for-quality-assurance/