Dateneffiziente Erkennung Neuer Fehlertypen in der Herstellung
Anmelder: Siemens Aktiengesellschaft 🇩🇪
Details
- Veröffentlichungs-Nr.
- EP4811061
- Anmeldetag
- 17. März 2025
- Veröffentlichung
- 23. September 2026
- Rechtsraum
- EP
- IPC
- G05B23/02
Abstract
The present disclosure provides a computer-implemented method for data-efficient detection of new failure types in manufacturing processes. The method includes defining a new failure type, labeling a minimal numbe2025000r of samples of the new failure type, and obtaining non-failure samples from a data pool without explicit labeling. Obtaining the non-failure samples comprises computing representations of the labeled samples, computing similarities between the representations of the labeled samples and representations of images in the image pool and selecting images from the image pool as non-failure samples based on the computed similarities. The method further includes fine-tuning a pre-trained model using the labeled samples of the new failure type, oversampling and the obtained non-failure samples to create a detection model for the new failure type.
Anmelder
- Firma
- Siemens Aktiengesellschaft
- Land
- 🇩🇪 Deutschland
Deutscher Technologiekonzern mit Sitz in München und Berlin. Aktiv in Automatisierungstechnik, Digitalisierung, Energie- und Bahntechnik sowie Industriesoftware für Industrie, Infrastruktur und Mobilität.
49.475 Patente in unserer Datenbank