Photovoltaic Panel Fault Detection Based on Improved Mask R-CNN
Tao Peng, Feng Wen · 2023
Solar photovoltaic panels are widely recognized as a clean energy generation device, and their quality and efficiency are becoming increasingly important for power generation. However, due to the harsh environmental conditions that PV panels are exposed to, such as high winds, heavy rain, and high temperatures, as well as damage during manufacturing and transportation, various types of failures can occur. These failures can lead to a reduction in the performance of the PV panels, affecting the efficiency and lifespan of the electricity generated. Therefore, it is crucial to detect and monitor PV panels for faults. The traditional target detection method has limitations, including low recognition accuracy and slow detection speed. This paper suggests an improved Mask R-CNN-based intelligent detection method for PV panel faults. The FPN in the Mask R-CNN model is improved to BiFPN to better reflect the original image information. In the later Mask stage, the channel attention mechanism is added to help the model focus more effectively and quickly on important areas of the image. Experiments show that the model can be applied to a variety of scenarios and effectively solves the problems of missed detection and false detection.