Machine learning-based battery electrode foil inspection
Ines Müller, Wenjing Song, Sebastian Georgi, Timo Eckhard, Alexander Korff
Kapitel/Beitrag aus dem Buch: Längle T. & Heizmann M. 2024. Forum Bildverarbeitung 2024.
Part of
Kapitel/Beitrag aus dem Buch: Längle T. & Heizmann M. 2024. Forum Bildverarbeitung 2024.
This paper presents an analysis of various autoencoder methods for automated anomaly detection. Prototype image datasets of battery foils, used as anode (copper foil) and cathode (aluminum foil) in lithium-ion batteries, are generated using a line-scan camera system with different illumination setups. The objective is to design and evaluate unsupervised learning methods for surface inspection of the foils. Additionally, the impact of different illumination geometries on the classification performance of the implemented models and their inference times is investigated and analyzed. Another objective is to accelerate model inference by integrating a DPU-based architecture, focusing on optimizing runtime performance for real-time anomaly detection. Using the DPU, an approach achieved a speedup by a factor of 40 compared to computations on the CPU.
Müller, I et al. 2024. Machine learning-based battery electrode foil inspection. In: Längle T. & Heizmann M (eds.), Forum Bildverarbeitung 2024. Karlsruhe: KIT Scientific Publishing. DOI: https://doi.org/10.58895/ksp/1000174496-6
This chapter distributed under the terms of the Creative Commons Attribution + ShareAlike 4.0 license. Copyright is retained by the author(s)
Dieses Buch ist Peer reviewed. Informationen dazu Hier finden Sie mehr Informationen zur wissenschaftlichen Qualitätssicherung der MAP-Publikationen.
Veröffentlicht am 21. November 2024