A Novel Approach for Efficient Solar Panel Fault Classification Using Coupled UDenseNet.

Area of Science:

• Renewable Energy Systems
• Artificial Intelligence in Engineering
• Materials Science

Background:

• Photovoltaic (PV) systems are vital for clean energy generation, but faults like shading and hot spots reduce efficiency and pose safety risks.
• Accurate PV fault classification is essential for optimal system performance and financial returns.
• Existing deep learning models often have high computational demands and struggle with complex image features and imbalanced datasets.

Purpose of the Study:

• To introduce a lightweight coupled UdenseNet model for enhanced PV fault classification.
• To improve the accuracy and efficiency of fault detection in photovoltaic systems.
• To address limitations of previous models in handling complex image features and imbalanced datasets.

Main Methods:

• Development and implementation of a lightweight coupled UdenseNet model.
• Utilizing geometric transformation and generative adversarial networks (GAN) for image augmentation.
• Training and validation of the model on diverse PV fault datasets.

Main Results:

• The UdenseNet model achieved high accuracy: 99.39% (2-class), 96.65% (11-class), and 95.72% (12-class).
• The model demonstrated improved efficiency with lower parameter counts, suitable for real-time analysis.
• Image augmentation techniques effectively enhanced performance on unbalanced datasets.

Conclusions:

• The proposed UdenseNet model offers a significant advancement in PV fault classification accuracy and efficiency.
• This lightweight model is well-suited for real-time monitoring of large-scale solar farms.
• The study highlights the importance of efficient AI models for the reliable operation of renewable energy infrastructure.