AI-supported surface inspection and measurement of aluminum ingots

A leading manufacturer of ingots (aluminum ingots) needs a reliable solution for 100% quality control. Surface defects impair further processing, while dimensional deviations lead to problems in production. In addition, different types must be reliably detected. Part of the image acquisition has to take place during movement, whereby a constant speed is crucial for precise inspection.

The solution consists of two inspection stations. In station 1, two high-resolution 2D cameras inspect the front and back of the aluminum ingots at a standstill. This involves a detailed analysis of surface defects such as air inclusions (blow holes) based on defined criteria. In station 2, four laser triangulation sensors record the height, width and shape of the aluminum ingots during movement. This process enables continuous quality control and the detection of shape deviations. The combination of rule-based image processing and deep learning algorithms (AI) enables precise defect classification. Our intuitive aku.visionManager® software interface allows the customer to easily customize the inspection criteria.

The implemented solution enables reliable detection of surface defects, precise geometry measurement and automated type testing. By integrating deep learning, the system recognises defect features even more accurately. Continuous detection during movement increases process speed and reduces downtimes. In addition, new product types can be flexibly taught in via the aku.visionManager® so that the system can also be adapted to future requirements.