Stimulation Initiative for European Neural Applications
Esprit Project 9811
Automatic Quality Control System for Tile-making Works
Company background
No information available
The problem
Until now, the classical testing of roofing tile's quality involved a
sound check for the detection of cracks and an inspection of the
surface. These test are usually run by sorters. In the past, attempts
for an automation of the process failed because of inadequate
measurement technology. The speed of inspection was limited to 45 tiles
per minute in manual testing.
Neural network application
The acoustic testing was automated by means of a neural network: The
trained neural network identifies the typical sound pattern of perfect
roofing tiles.
Benefits
By using neural networks in acoustic testing even hardly perceivable
hairline cracks can be detected, which remained undiscovered in the
customary manual procedures.
Generalization
This example shows that industrial quality control can be considered as
a vast area for neural network application.The testing of product's
quality is a typical problem in industrial production for which very
often no general method exists. In most cases, the testing is based on
the experience of sorters. The application of neural networks in
industrial quality control can help to improve the quality of the
products and simplify the process of control.
Contact Person
Medav Digitale Signalverarbeitung GmbH -
Dr. Hans-Joachim Kolb.
Gräfenberger Str.34,
D-91080 Uttenreuth.
Tel.: +49 9131 583 10,
Fax: +49 9131 583 11.
![[ ]](../logo_small.gif){kind=logo}
SIENA