Photovoltaic Panel Intelligent Detection Method Based on Improved Faster-RCNN

The distribution environment of large-scale photovoltaic power plants is complex, and the operation and maintenance of photovoltaic modules in the future cannot rely on manual inspection. However, there are problems such as poor accuracy and low efficiency of traditional target detection in the current UAV (Unmanned Aerial Vehicle) inspection work, which cannot meet the necessary operation and maintenance. Therefore, this paper proposes an intelligent detection method for photovoltaic power panels based on the improved Faster-RCNN target detection algorithm to analyze and identify images taken during UAV inspection. First, photovoltaic module images are collected by UAV equipped with infrared thermal imaging cameras. Next, the collected PV module defects are labeled. Finally, the improved Faster R-CNN target detection algorithm is used to build a deep learning network model to classify fault types accurately. It can solve the problems such as insufficient feature extraction of CNN network, missed detection and false detection of photovoltaic energy panels with uneven light distribution. The results show that the average accuracy of the proposed algorithm on the photovoltaic module dataset reaches 95.9%.