Advanced fault detection in PV panels using deep neural networks: leveraging transfer learning and electroluminescence image processing
An ML framework detects faults in PV panels from their unlabeled EL images using transfer learning, k -means labeling, and data augmentation. CNN models achieve 98–99.5% accuracy for binary classification, reducing manual labeling and improving PV system reliability.
Authors
- Hafiz Imtiaz (ORCID: https://orcid.org/0000-0002-2042-5941)
- Ihtesham Ibn Malek
Institutions
- Bangladesh University of Engineering and Technology (BD)
Publication Details
- Journal
- Energy Advances
- Published
- 2025-12-09
- DOI
- https://doi.org/10.1039/d5ya00239g
- Primary Topic
- Photovoltaic System Optimization Techniques
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- University of Engineering and Technology, Lahore
- Bangladesh University of Engineering and Technology