Photovoltaic Module Infrared Image Fault Detection System Major:Electronic Information

Photovoltaic modules are prone to hot spot faults during power generation due to aging and harsh environmental factors.Regular hot spot detection can ensure the high-efficiency operation of photovoltaic power plants,reduce component damage,and prevent fires.Traditional manual inspection methods have low detection efficiency,high difficulty,and high cost,which is not conducive to the long-term development of photovoltaic power plants.Currently,the photovoltaic hot spot detection method has low detection accuracy and high missed detection rate,which is difficult to meet the actual needs of photovoltaic power plants.Therefore,this article focuses on the infrared image of photovoltaic modules and researches a fault recognition method based on deep learning.The main work includes:(1)To address the problem of the photovoltaic infrared image having a relatively simple feature and high similarity between features,making it difficult to extract hot spot features,a hot spot detection model based on Transformer was studied.Swin Transformer was used as the feature extraction network in the Faster R-CNN model to capture the global information of the image and establish dependencies between features to improve the modeling ability of the model.Comparative experiments on the photovoltaic infrared image dataset demonstrated that using Swin Transformer as the feature extraction network in Faster R-CNN can effectively improve the ability to extract hot spot features.(2)Aiming at the characteristics of small and indistinctive targets in hot spot faults,a feature pyramid network is utilized for feature fusion to improve the problem of hot spot faults being easily ignored by models due to small targets and indistinctive features.At the same time,to suppress interference such as background and noise in photovoltaic infrared images,a lightweight attention module is introduced to make the model focus on important channels and key areas,improving the detection accuracy of hot spot faults.Through comparative experiments and ablation experiments on photovoltaic infrared image datasets,the hot spot identification method based on feature fusion and attention mechanism is verified to be effective in detecting hot spot faults on photovoltaic components.This model has achieved high improvements in both accuracy and recall rate.(3)This article conducts a requirement analysis of the photovoltaic module fault detection system,determines the framework and technology used to implement the system,designs and implements various functional modules,databases,and visual interfaces,and builds a photovoltaic module fault detection system.Through functional and non functional testing of the system,it was verified that the system built in this paper can run stably and meet the detection requirements of photovoltaic power plants.