Related Experiment Video
Updated: Sep 16, 2025
11:06
Fabrication of Fully Solution Processed Inorganic Nanocrystal Photovoltaic Devices
Published on: July 8, 2016
10.6K
ResNet-based image processing approach for precise detection of cracks in photovoltaic panels
Montaser Abdelsattar1, Ahmed AbdelMoety2, Ahmed Emad-Eldeen3
1Electrical Engineering Department, Faculty of Engineering, South Valley University, Qena, 83523, Egypt. Montaser.A.Elsattar@eng.svu.edu.eg.
Scientific Reports
|July 8, 2025
Summary
A new Deep Learning (DL) method using Residual Network (ResNet) accurately detects cracks in solar Photovoltaic (PV) panels from Electroluminescence (EL) images. ResNet34 offers the best balance of performance and efficiency for PV module maintenance.
Area of Science:
- Renewable Energy
- Materials Science
- Computer Science
Background:
- Advancing renewable energy necessitates durable and efficient solar Photovoltaic (PV) modules.
- Early detection of defects like cracks is crucial for maintaining PV panel performance and longevity.
- Existing methods may lack the accuracy or efficiency needed for large-scale PV system maintenance.
Purpose of the Study:
- To propose a novel Deep Learning (DL) mechanism for accurate crack detection in PV panels using Electroluminescence (EL) images.
- To evaluate different Residual Network (ResNet) architectures for optimal performance and computational efficiency.
- To develop an efficient crack detection system for enhanced PV panel maintenance and solar energy sustainability.
Main Methods:
- Utilized a large dataset of 2000 EL images from polycrystalline and monocrystalline PV cells.
- Implemented and tested various ResNet architectures (ResNet34, ResNet50, ResNet152) for crack detection.
- Split the dataset into training (70%), validation (20%), and testing (10%) sets for model evaluation.
Main Results:
- ResNet architectures achieved high F1-Scores: ResNet34 (86.63%), ResNet50 (87.37%), and ResNet152 (88.89%).
- ResNet34 was selected as the optimal model due to its superior trade-off between detection accuracy and computational performance.
- The developed system demonstrates effective early defect diagnosis from raw EL image data.
Conclusions:
- The proposed DL-based crack detection system enhances PV panel maintenance, bolstering solar energy sustainability.
- This research provides a practical solution for the renewable energy sector, especially with new materials in PV panels.
- The study establishes a foundation for advanced image-based defect detection methods in PV systems, preserving module efficiency.
