Synthetic Data-Enhanced Microcrack Detection in Photovoltaic (PV) Cells Using YOLO11 for Smart Industry Applications

The recent advancements in smart industry and advanced automation are reshaping industries by introducing cutting-edge technologies that improve efficiency, sustainability, and productivity across sectors such as manufacturing, energy, and infrastructure. In particular, the application of automated systems for monitoring and maintenance tasks plays a crucial role in the optimization of industrial operations. One such area is the monitoring of photovoltaic (PV) cells, which are integral components of renewable energy systems. Detecting defects, such as microcracks, in these cells is critical for ensuring their efficiency and longevity. This study focuses on the use of object detection algorithms, specifically YOLO11, to identify and classify microcracks in PV cells. A key innovation in this research is the use of both real and synthetic crack images in training the detection model. By artificially creating cracks of varying sizes, orientations, and locations on PV cells, we expand the training dataset, improving the model's robustness and accuracy in detecting defects. The study also compares the performance of YOLO11 with previous versions of the YOLO algorithm, YOLOv8, YOLOv9, and YOLOv10, across multiple performance metrics, using both primary (real crack) and expanded (synthetic crack) datasets. The results demonstrate that YOLO11 achieves high precision and recall values, particularly for detecting large cracks, with notable improvements when the expanded dataset is utilized. The proposed work highlights the potential of advanced object detection algorithms in the context of smart industry and advanced automation, offering a scalable solution for real-time monitoring and maintenance of PV cells.