Photovoltaic performance comprehensive monitoring evaluation and maintenance method and system

The invention discloses a photovoltaic performance comprehensive monitoring evaluation and maintenance method and system. According to the method, the deep features and the shallow features of the photovoltaic module are obtained through the YOLO model, the VGG neural network model and the Faster R-CNN neural network, so that the accuracy of a monitoring result is effectively improved; meanwhile, the corresponding photoelectric conversion efficiency is obtained through the I-V output characteristics corresponding to the photovoltaic cell units, and whether the reason for low efficiency of the photovoltaic cell units is that dirt exists in shells of the photovoltaic cell units or not is rapidly judged in combination with an image monitoring result, the photoelectric conversion efficiency and an angle recognition result; the system is simple in structure, image features and photoelectric parameters of the photovoltaic system are utilized, the performance of the photovoltaic system can be comprehensively monitor