Product

Industries

Resources

Company

Surface Inspection

Inspection of sealing membranes

Inspection of sealing membranes

By combining an in-line inspection device with line-scan cameras and the Maddox AI software, sealing sheets are reliably and 100% inspected – even at high web speeds.

Product

Product

Sealing Membranes

Industry

Industry

Automotive

Total savings / year​

Total savings / year​

€300k–500k

A black web of material runs over rollers through an industrial production facility.

Customer Request

Customer Request

Our customer manufactures sealing sheets as continuous material in various product variants, which were previously inspected manually by employees during the ongoing process. Although the inspectors worked conscientiously, subjectivity and fatigue led to fluctuating results: OK goods were sometimes sorted out as NOK, while NOK goods passed through unnoticed. The consequences were increased scrap and a risk of customer complaints. The dark, textured, and partly reflective surface of the sheets, as well as the high product variability, made the use of classic, rule-based inspection systems unreliable. Maddox AI has developed an inspection device that is integrated directly into the existing production line and captures 100% of the moving sheet from above. A rotary encoder triggers the cameras with precise timing, the captured images are processed locally on the edge device, and the inspection result (OK/NOK) is immediately transferred to the existing PLC. Detected defects include tears, holes, wrinkles, melted spots, bubbles, foreign objects, smears, and missing material – across multiple product variants. Changeover between variants is automatic: the operator simply selects the desired article number on the touchscreen.

Before

Before

Previous Inspection System

Previous Inspection System

Previous Inspection System

The webs were manually inspected during the ongoing process. Detection accuracy varied depending on the shift, inspector, and daily performance.

Increased scrap due to Pseudo scrap. Non-defective parts were sorted out as a precaution, isolated genuine defects reached the customer.

Since the test results were not digitized, it was not possible to conduct a structured root cause analysis for the occurrence of non-conforming goods.

After

After

With Maddox AI

With Maddox AI

With Maddox AI

The webs were manually inspected during the ongoing process. Detection accuracy varied depending on the shift, inspector, and daily performance.

The entire web is inspected 100% automatically according to a precise definition of defects – as accurately as a constantly attentive human inspector. Operation is carried out by one employee per shift.

Thanks to the reliable distinction between acceptable irregularities and real defects, both waste and the risk of complaints are significantly reduced.

Through digitized error analysis and the defect heatmap, structurally occurring errors can be identified and process optimizations can be implemented in a targeted manner.

Increased scrap due to Pseudo scrap. Non-defective parts were sorted out as a precaution, isolated genuine defects reached the customer.

Thanks to the reliable distinction between acceptable irregularities and real defects, both waste and the risk of complaints are significantly reduced.

Since the test results were not digitized, it was not possible to conduct a structured root cause analysis for the occurrence of non-conforming goods.

Through digitized error analysis and the defect heatmap, structurally occurring errors can be identified and process optimizations can be implemented in a targeted manner.

90%

90%

90%

less manual sorting costs due to less Pseudo scrap

50 m/min.

50 m/min.

50 m/min.

Inspection speed

100%

100%

100%

MES/ERP integration via REST API

Want to learn more about our solution?

Try it free now!

Would you like to learn more about our solution?

Try it now for free!

Would you like to learn more about our solution?

Try it now for free!