Thermal spot fault detection of photovoltaic modules using improved YOLOv5 algorithm

The thermal spot fault of photovoltaic modules is a key problem that affects the power generation efficiency and safe operation of the power station, but in the detection process containing small target thermal spot fault, it is easy to lose information, false detection or missing detection. To solve the problem of missing detection of small target and poor adaptability of multi-scale, this paper proposed a photovoltaic thermal spot detection algorithm using improved YOLOv5. This paper proposed a photovoltaic thermal spot detection model based on improved YOLOv5 algorithm. The algorithm is an improved version of YOLOv5, wherein P2 detection layer is added to enhance the detection accuracy of small targets. Moreover, the GD(Gather-and-Distribute) module of Gold-YOLO model is applied to improve the feature fusion module of traditional YOLOv5. The model achieved a F1Score of 0.77 at a confidence threshold of 0.403, and a mean average precision (mAP) of 0.762 at an IoU threshold of 0.5, which are 4.53% higher than that of the original algorithm.The experimental results show that the proposed algorithm can effectively improve the reliability of the system. Thus, this algorithm shows considerable potential in small target photovoltaic thermal spot fault detection.