.png){kind=icon}
.png){kind=icon}
Return
Diagnostics in multi-line photovoltaic module manufacturing using electroluminescence-based unsupervised anomaly detection
.png){kind=icon}
PRE
AI
X
H
Z
DOI：10.1016/j.solener.2026.114674.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• We first formulate photovoltaic multi-line manufacturing diagnostics as a multi-class unsupervised anomaly detection. • We propose UniDisc with DFE and DFS for robust diagnostics. • We improve ELAD diagnostic image-level AUROC by 1.8% over the second-best. • We validate image/pixel metrics on ELAD, showing strong diagnostic and localization advantages.
Keywords:
photovoltaic manufacturing
unsupervised anomaly detection
electroluminescence imaging
diagnostic metrics
anomaly localization
Journal
IF:
6.6
Papers:
1.4W
Citations:
6.2W
Organization
No organization information available
Cited Papers