Research on Photovoltaic Substation Meter Reading Identification Method Based on Improved FasterRCNN

The digital meter of photovoltaic substation is an important equipment for photovoltaic power generation management. In response to the problem of video reading of digital data, this paper proposes an improved object detection algorithm Faster R-CNN to solve this problem based on the requirements of the scientific research project currently being implemented in Changqing Oilfield. The main content is to modify the backbone network from VGG16 to ResNet50 to enhance feature extraction capabilities; In addition, to solve the problem of difficult detection of large readings caused by close shooting distances, an improved FPN is used as an additional module to the backbone. Finally, a Convolutional Block Attention Module (CBAM) is introduced to enhance the detection ability of small targets and further enhance feature representation and network perception. The experimental results showed that each improvement had varying degrees of improvement, and after adding the improvement module, the mAP reached 86.5%, an increase of 7.1% compared to before the improvement.