Detection Of Surface Obstructions Of Photovoltaic Modules Based On Polarization Imaging

Photovoltaic(PV)power generation is widely promoted and applied as an environmentally friendly and economical power generation method.PV modules are important devices for achieving photoelectric conversion in PV systems.The working environment is relatively harsh,and the surface often appears with dust,fallen leaves,bird droppings,and other obstructions,which seriously reduces the power generation efficiency of PV modules,reduces their service life,and may generate heat spot effects that cause fires,causing property and economic losses.Therefore,it is particularly important to accurately detect and timely clean the surface obstructions of PV modules.The most commonly used detection method is the measurement of electrical parameters,which is difficult to accurately locate and classify obstructions.In recent years,with the development of visual technology,target detection based on visible light images has been gradually applied to obstructions on the surface of PV modules and has made significant progress.However,its detection accuracy still needs to be improved.Visible light images do not fully consider the characteristics of the obstructions and the differences in characteristics between the obstructions and the PV module;When collecting an obstruction image under insufficient lighting conditions on overcast days or strong reflection of PV modules on sunny days,there may be a problem of unclear shooting,which reduces detection accuracy.Polarization is an inherent attribute of light,and there are significant differences in polarization characteristics between the obstruction and the PV module.Therefore,using polarization imaging to obtain polarization information of a target can improve the discrimination between the two.In this paper,we carry out research on polarization imaging-based detection technology for dust accumulation and fallen leaves on the surface of PV modules,and the main work is as follows:(1)Basic theory.Firstly,the basic theory of polarized optics and the representation method of polarized light are introduced.Then,the basic structure of PV modules is studied,and the physical characteristics of typical obstructions such as fallen leaves and dust are discussed.On this basis,the impact of the basic structure of PV modules and the types of obstructions on the polarization characteristics of reflected radiation is analyzed,providing a theoretical basis for the effective detection of surface obstructions on PV modules in the future.(2)A detection model for dust accumulation density of PV modules based on polarization reflection characteristics is proposed.Firstly,the interaction mechanism between visible light and dust particles is studied,and the optical polarization scattering model of dust particles on the surface of PV modules under visible light irradiation is presented.Then,according to Fresnel’s law,the transmission process of polarized reflected radiation of light rays on the surface of dust accumulation PV modules is derived,and the physical mechanism of polarization information,the essential characteristics of a dust accumulation reflected radiation,is clarified.On this basis,an observation platform for dust accumulation density is established,and dust accumulation samples are prepared and naturally deposited on the surface of PV modules.Polarized reflection images of dust accumulation PV modules are obtained.Functional relationships between normalized reflected radiation intensity,degree of polarization(DOP),and dust accumulation density are established.Dust accumulation density detection models are established and verified through indoor and outdoor experiments.The results show that the reflected radiation intensity increases with the increase of dust accumulation density,and is greatly affected by irradiance and incident angle.The DOP decreases with the increase of dust accumulation density,and the DOP decreases rapidly at the beginning of ash deposition,both of which have a good correlation with dust accumulation density;The detection model has high accuracy,with an average relative error of less than 6.50%.(3)A model for detecting surface fallen leaves of PV modules based on improved YOLOv5 is proposed.Firstly,according to the impact of fallen leaves on the polarization characteristics of PV modules,analyze the characteristics of the DOP image,and modify the YOLOv5 model from three aspects: feature extraction network,coordinate attention mechanism,and loss function.Then,by building a detection platform to obtain fallen leaves polarization images and expanding and dividing the dataset,the improved model is evaluated using accuracy,recall,and average accuracy as the main evaluation indicators.The results show that the improved YOLOv5 model can obtain more shallow information of fallen leaves in the DOP image,make the edge outline of fallen leaves more clear in the feature map,suppress irrelevant information such as background,improve the contrast between fallen leaves and PV modules in the feature map,and have a good detection effect.Figure [54] table [19] reference [80]...