Research On Calculation Method Of Photovoltaic Power Based On Ground-based Cloud Image

Photovoltaic power is affected by meteorological and environmental factors,so it has volatility and randomness,which brings challenges to photovoltaic system management and power grid operation.Therefore,it is necessary to predict the photovoltaic power.With the successful development of sky observation devices,it is possible to use ground-based cloud images combined with image processing to predict photovoltaic power in a short time.Photovoltaic power prediction based on ground-based cloud image can be understood as the calculation of photovoltaic power of ground-based cloud image in the future,so the calculation of photovoltaic power is an important part of photovoltaic power prediction.The accuracy of photovoltaic power prediction based on ground-based cloud image mainly depends on the image processing method and the reflection degree of extracted features to photovoltaic power influencing factors.In this thesis,the cloud image preprocessing and feature extraction is performed on the cloud image collected by the all-sky imager,and a reliable ground-based photovoltaic power calculation model is established.The main work of this thesis is as follows:(1)Ground-based cloud image preprocessing.The effective region of the ground-based cloud image is extracted,and according to the characteristics of the cloud images,the template calibration method is used to correct the distortion of the images and the bilinear interpolation method is used to fill in the missing value,so as to complete the missing information and ensure the integrity of the cloud images.The improved Retinex algorithm is used for illumination correction of cloud images to ensure the uniformity of cloud color and texture,which establishes the basis for subsequent feature extraction.(2)Cloud recognition of ground-based cloud image.According to the characteristics of the cloud images,the gray scale of the cloud image is carried out.By analyzing the characteristics of the clear sky dictionary and data set,an improved image segmentation method based on the clear sky dictionary is proposed to identify cloud clusters.(3)Feature extraction of ground-based cloud image.The factors that affect the change of photovoltaic power are analyzed.Based on the cloud image and geographic information,the features related to the change factors are extracted,and the ground-based cloud image feature database is established.The validity of extracted features is verified by the correlation analysis method.(4)Calculation of photovoltaic power.The random forest algorithm is used to establish the photovoltaic power regression model based on the features extracted from the ground-based cloud images and the photovoltaic power data,so as to realize the photovoltaic power calculation based on the cloud image.