Karriere am WZL


Masterarbeit oder Bachelorarbeit oder Projektarbeit

am Forschungsbereich Fertigungsmesstechnik und Qualitätsmanagement, Abteilung Model-based Systems, Gruppe Industrial Tomography

Development of a machine learning approach for depth determination of defects in CFRP parts

CFRP have become an interesting alternative to metals in many industries, such as the automotive sector or gas and oil industries, due to their mechanical properties and their low weight. Current research concentrates on the reliable detection and classification of potentially severe defects that reduce the integri-ty of CFRP components. In our research group, we are developing techniques that estimate the depth of these defects using active lock-in thermography. Thereby, depth information is retrieved from the phase of reflected heat waves by comparison with the expected phase provided by a physical 2D wave field model.

The task is to improve the performance of the developed approach by replacing the physical model, which just provides a rough estimation of the complex physical situation of the heat transfer in CFRP parts, for a black box model.
For this purpose, calibrated parts with known defects are used and scanned with a CT, that provides a benchmark measurement in this case. Using the thermography data and the known depths from the CT scan, a model is developed by supervised training of a neural network. The approach is validated by applying the trained model on thermography data of similar CFRP parts with known defects.

We offer:
• Interesting tasks
• Extensive supervision
• Independent conduction of an individual research project
• Working in an international research community (Mainly Brazil)

• Student of physics, computer science, engineering, etc.
• Interest in programming, image processing and machine learning
• Knowledge of one programming language (Preferentially Python)
• English skills (research is conducted in strong collaboration with scientist of UFSC, Brazil)
Zeitaufwand: 300,00 Arbeitsstunden

Dominik Wolfschläger
Cluster Produktionstechnik 3A 146
Tel.: +49 241 80-27337
Fax: +49 241 80-22293
Mail: D.Wolfschlaeger@wzl.rwth-aachen.de