Anomaly Detection for Circuit Waveforms Using Diffusion Model
IEEE Transactions on Consumer Electronics · 2026-05 · 72 (2) · 3261-3272
DOI: 10.1109/tce.2026.3678259 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Consumer Electronics, 2026-05. Verfasst von Hung-Wei Hsueh und 4 weiteren Autor:innen. Die Arbeit nennt 44 Literaturangaben. Gefördert von: Google Research Award, National Science and Technology Council, Taiwan.
Abstract (Originalsprache)
Circuit waveform inspection is critical to ensuring performance and reliability in consumer electronic products. However, engineers in industrial production lines often inspect waveforms manually, requiring technical expertise and time. To automate and speed up testing, we propose a diffusion model based normal waveform generator (DNWG) that reconstructs waveform images using only normal data, eliminating abnormal sample collection and retraining for unknown anomalies. By integrating masking techniques to preserve output accuracy, DNWG enables anomaly detection for previously unseen issues. In addition, we introduce the compositional similarity index, which considers the electrical and physical properties of circuits, improving upon conventional similarity metrics. This approach increases efficiency for anomaly detection in circuit waveform inspection.
Förderung
- Google Research Award
- National Science and Technology Council, Taiwan
Zitieren
APA
Hsueh, H., Lee, C., Hsu, H., Tseng, C., & Hsu, S. (2026). Anomaly Detection for Circuit Waveforms Using Diffusion Model. IEEE Transactions on Consumer Electronics, 72(2), 3261-3272. https://doi.org/10.1109/tce.2026.3678259
Vancouver
Hsueh H, Lee C, Hsu H, Tseng C, Hsu S. Anomaly Detection for Circuit Waveforms Using Diffusion Model. IEEE Transactions on Consumer Electronics. 2026;72(2):3261-3272. doi:10.1109/tce.2026.3678259
BibTeX
@article{Hsueh2026,
author = {Hsueh, Hung-Wei and Lee, Chia-Hsun and Hsu, Hao-Ming and Tseng, Ching-Wei and Hsu, Shu-Han},
title = {Anomaly Detection for Circuit Waveforms Using Diffusion Model},
journal = {IEEE Transactions on Consumer Electronics},
year = {2026},
volume = {72},
number = {2},
pages = {3261--3272},
doi = {10.1109/tce.2026.3678259}
}
EUVIMED