28-04-2023 дата публикации
Номер: CN116030409A
Принадлежит:
The invention relates to the technical field of power photovoltaic panels, and particularly discloses a photovoltaic panel dust deposition state recognition method based on adaptive image segmentation, which comprises the following steps: S1, image preprocessing; s2, photovoltaic string segmentation is carried out; s3, dust cutting; s4, partitioning the image; s5, feature extraction; s6, identifying a dust state; according to the method, the collected image and the segmented image are both color images, so that the segmented image is subjected to block processing in order to improve the algorithm operation speed, the operation speed can be improved, the number of image texture features can be increased, feature extraction is performed on dust by using the LBP algorithm, and the feature extraction efficiency is improved. And finally, state identification is carried out by a softmax classifier. Through identification of the dust deposition state of the photovoltaic panel, hot spots and breakage ...
Подробнее