AI-Driven Quality Assurance in Pressing

HUHN Pauli Deutschland GmbH specializes in high-speed cold forming of metals using progressive dies. Quality defects must be detected early, as each defective part results in waste and lost production capacity. Previously, only statistical spot checks were performed.

aiXbrain analyzes process data from the tooling with AI to identify emerging quality defects and relieve heavily utilized operators.

• Integration with sensor platform
• Classification of part quality with AI
• Prevention of scrap and machine downtime

As part of the BMBF-funded joint project GeMeKI, a progressive die was equipped with a specialized sensor package and the meastream Industrial Edge Toolbox. The AI operates on pressure and vibration signals that are spectrally transformed per press stroke (FFT).

Through a targeted measurement campaign, training data was generated to cover even rare wear effects. Supervised deep learning and unsupervised random forest models are used for training.

The trained AI precisely detects bending angle quality and distinguishes defects such as too-small or too-large bending angles — enabling cost-effective in-situ quality monitoring.

The combination of sensor package, edge computing, and AI enables 100% quality control, reducing scrap and unplanned maintenance while relieving employees.

In perspective, savings of five to ten percent of manufacturing costs are achievable. aiXbrain offers the AI component via the Dataray® AI Framework as a managed service or AI app in a SaaS model.