Research On Hot Spot Fault Detection Of Photovoltaic Modules

With the massive development and utilization of fossil energy,environmental pollution has become a global problem,and the development and utilization of new and renewable energy is imminent.Solar energy plays an important role in the field of new energy power generation with the characteristics of no consumption of resources,no pollution and inexhaustible use.However,as the core device of the photovoltaic power generation system,any fault problem generated during the operation will affect the normal operation of the entire photovoltaic power generation system,so it is very necessary to study the fault problem of the photovoltaic module.Hot spot fault is a common fault problem during the operation of photovoltaic modules.The detection of hot spot fault based on infrared image has the advantages of high detection efficiency,low use cost,and no damage,and has become a common hot spot fault detection method.This paper takes the infrared image of photovoltaic hot spot as the research object,completes the denoising and segmentation of the infrared image of photovoltaic hot spot,and finally realizes the fault detection of photovoltaic hot spot.The specific research contents are as follows:1.A hybrid noise adaptive denoising algorithm based on principal component analysis is proposed.In view of the characteristics of the hot spot infrared image with Gaussian and salt and pepper mixed noise,an adaptive denoising algorithm based on principal component analysis is proposed.Before principal component extraction,an adaptive window preprocessing is performed to filter the high-density salt and pepper noise in the image,and then the information component is extracted based on the local similarity of the image,the noise component is filtered,and the noise level is updated for secondary denoising,It improves the denoising performance of the algorithm for high-density noise.The experimental results show that this algorithm has a good denoising effect for the mixed noise of the hot spot image,and the contour of the hot spot area is obvious.2.A photovoltaic infrared image segmentation method based on improved U-net network is proposed.Aiming at the problem of over-segmentation and undersegmentation of traditional image segmentation methods in complex gray images,an improved U-net network image segmentation method is proposed.Firstly,the gray level co-occurrence matrix and the original image information are synchronously input into the U-net network to improve the perception ability of the U-net network to the image texture features;Then the idea of multi-scale feature fusion is introduced into the encoding and decoding structure of U-net network,which enables the network to extract more abundant image features;Finally,the use of depth can separate the convolution,reduce the parameters required for network training,and improve the running speed and generalization ability of the network model.Experiments show that the segmentation results of this algorithm are accurate and robust.3.A fault detection model based on improved YOLOX algorithm is proposed.The YOLOX-S model with low parameter quantity and high robustness is selected as the main network structure.The weighted bidirectional feature pyramid network is used in the enhanced feature extraction network part.The improved YOLOX-S network model is trained and tested with the original YOLOX-S and YOLOX.The comparison results show that the improved model has a high m AP value,which ensures the accuracy of fault detection while reducing the operation cost,and has practical application value.Figure [48] Table [7] Reference [80]...