Sign in Subscribe
Artificial neural networks

Automated Detection of Shading Faults in Photovoltaic Modules Using Convolutional Neural Networks

Strategic Insights

— 1 min read

Researchers from the Autonomous University Queretaro in Mexico have made significant breakthroughs in the field of photovoltaic technology by developing a novel approach to detect and classify shading faults in photovoltaic modules using convolutional neural networks. This innovative method has the potential to mitigate climate change and advance sustainable development by enhancing the efficiency and reliability of renewable energy systems.

Key Takeaways:

  • The researchers used convolutional neural networks (CNNs) to automatically detect and classify shading faults in photovoltaic modules, achieving classification accuracies of 99.5% under noiseless conditions and 90.1% under a 10 dB noise condition.
  • The study simulated four scenarios in Simulink: a healthy module and three levels of shading severity (light, moderate, and severe), and transformed the resulting I-V curves into grayscale images to train and evaluate custom-designed CNN architectures.
  • The researchers tested multiple network configurations, varying image resolution, network depth, and filter parameters, to explore their impact on classification accuracy and robustness.
  • The study found that CNN-based approaches can be both effective and computationally lightweight, making them a promising solution for integration into automated diagnostic tools for PV systems.

Statistics:

  • Classification accuracy under noiseless conditions: 99.5%
  • Classification accuracy under 10 dB noise condition: 90.1%
  • Number of simulated scenarios: 4
  • Levels of shading severity simulated: 3 (light, moderate, and severe)
  • Number of network configurations tested: multiple

Sources:

  • Processes, 2025;13(9):2999
  • NewsRx. New Data from Autonomous University Queretaro Illuminate Findings in Technology (Automated Detection of Shading Faults In Photovoltaic Modules Using Convolutional Neural Networks and I-v Curves). Journal of Engineering. October 20, 2025; p 1598.

Note: The provided text was used as the sole source of information, and all data and statistics were extracted directly from the text.

Read more