ET | INF | MT | PHY | Thesis | SHK |WHK
Motivation:
Laser material processing is increasingly important in industrial
manufacturing due to its precision and efficiency. In deep penetration
welding, a vapor capillary („keyhole“) forms and moves along the weld path.
Its stability directly affects weld quality. However, materials like aluminum
castings and copper alloys often cause keyhole fluctuations due to volatile
elements, making real-time monitoring essential.
Building on previous work, we plan to combine digital holography and AIbased data analysis for real-time 3D imaging. This is a transfer project
together with the Fraunhofer IWS and the industry partners Lessmüller
Lasertechnik GmbH and ZF Group.
Keywords digital holography, deep learning, image processing
Tasks (software):
- Implementation of neural network based reconstruction and classification
- Model optimization, result evaluation and documentation
- Developing a method to model reconstruction uncertainty and quantify confidence
Tasks (hands on): Characterization and optimisation of lab setup
Contact
Alexander Thees alexander.thees@tu-dresden.de