Inline Quality Assessment of Hairpin Laser Welds in Traction Drives Using 3D Laser Triangulation and 3D Deep Learning

Zorila I, Baader M, Schatzl M, Franke J, Risch F (2026)

Publication Type: Conference contribution

Publication year: 2026

Original Authors: Iulian Zorila, Marcel Baader, Markus Schatzl, Jörg Franke, Florian Risch

Event location: Karlsruhe DE

DOI: 10.1109/EPTS67931.2025.11547612

Abstract

The electrification of vehicles poses significant challenges to automotive manufacturers, particularly in the production of electric traction drives. Winding concepts such as Hairpin, I-pin or X-pin technology are increasingly relying on the use of rectangular copper wires instead of round wires for the stator windings of the motor. This requires the laser welding of a large number of wire pairs to connect individual winding sections. The quality of the joints is critical to the overall functionality and product reliability. To ensure their functional integrity, suitable sensor technology is increasingly being pursued to monitor and check the process. This work focuses on whether the quality of the welds can be adequately assessed using 3D laser triangulation sensors. A dual 3D sensor device was developed to capture high-resolution 3D point clouds, including vertical pin areas, in a fast inline process. The resulting topology representation accurately preserves surface features and defects. Evaluation uses a 3D deep learning approach incorporating production-derived factors. Results show laser triangulation sensors can generate high-resolution 3D representations, enhancing neural network learning efficiency compared to 2D projection approaches. The system enables nondestructive, inline assessment of welding quality in electric motor production. To the best of the authors knowledge, genuine 3D deep learning had not been shown applicable in inline stator manufacturing in real industrial settings. This research thus extends application potential for such technologies and advances the field of 3D deep learning.

Authors with CRIS profile

Marcel Baader Lehrstuhl für Fertigungsautomatisierung und Produktionssystematik (FAPS) Jörg Franke Lehrstuhl für Fertigungsautomatisierung und Produktionssystematik (FAPS) Florian Risch Lehrstuhl für Fertigungsautomatisierung und Produktionssystematik (FAPS)

Involved external institutions

senswork GmbH - Innovation Lab DE Germany (DE) senswork GmbH - Innovation Lab DE Germany (DE)

How to cite

APA:

Zorila, I., Baader, M., Schatzl, M., Franke, J., & Risch, F. (2025). Inline Quality Assessment of Hairpin Laser Welds in Traction Drives Using 3D Laser Triangulation and 3D Deep Learning. In Proceedings of the 2025 2nd International Conference on Production Technologies and Systems for E-Mobility (EPTS). Karlsruhe, DE.

MLA:

Zorila, Iulian, et al. "Inline Quality Assessment of Hairpin Laser Welds in Traction Drives Using 3D Laser Triangulation and 3D Deep Learning." Proceedings of the 2025 2nd International Conference on Production Technologies and Systems for E-Mobility (EPTS), Karlsruhe 2025.

BibTeX: Download