This study introduces an ensemble artificial intelligence model that combines multiple neural network approaches to detect and classify seven types of PV faults, supporting more reliable and efficient solar energy systems.
Key findings:
- Faults in PV arrays, including ageing, shading and short circuits, can reduce power generation and create operational risks.
- The proposed model combines three AI approaches, DBN-ELM, CNN-SVM and CNN-BiGRU, using weighted probability averaging to improve fault detection.
- The ensemble model achieves 99.0% overall accuracy, demonstrating strong performance in identifying both single and multiple faults.
This article was published in the WSSET affiliate journal, Energy Catalyst.
For more information on our affiliate journals, visit: wsset.org/journals
Read the full article here: https://caravelpress.com/journals/ec/articles/2026.006