| In recent years,the photovoltaic industry has developed rapidly,and photovoltaic power station has become one of the main sources of renewable energy power generation.Photovoltaic cells are one of the most critical components in photovoltaic modules.Its quality directly affects the power generation efficiency and operation status of photovoltaic modules.Due to the characteristics of the crystal structure,photovoltaic cells are prone to defects such as crack,scratch and broken gate in the production process.With the development of photovoltaic cells in the direction of thinner and thinner,the probability of defects is gradually increasing.Therefore,in order to ensure the performance and health of photovoltaic cells,it is very important to detect their defects effectively.Infrared thermography nondestructive testing technology has the advantages of non-contact,intuitive and efficient,and has been successfully applied to the qualitative characterization and quantitative analysis of photovoltaic cells.Aiming at the defects of photovoltaic cells,this paper uses Electrical Pulsed Infrared Thermography（EPIT）technology to detect the defects of photovoltaic cells from the perspective of rapid detection and intelligent recognition at the factory end,and analyzes the heat dissipation and transmission mechanism of photovoltaic cells and the theory of infrared thermal wave heat conduction.The EPIT system was built;the research on photovoltaic cells defect detection,infrared thermography sequence processing and analysis,and automatic classification and detection of photovoltaic cell defects were carried out.Firstly,the heat dissipation and transmission mechanism and heat conduction process of photovoltaic cells are analyzed.The experimental system of Electrical Pulsed Infrared Thermography（EPIT）is built,and the composition and main performance parameters of the experimental system are briefly analyzed.The photovoltaic cell test samples were introduced,and the experimental parameters were set.The infrared thermal wave signal of polycrystalline silicon photovoltaic cell defect samples surface was collected,and the temperature information of photovoltaic cell samples surface was analyzed by experimental research,which proved the detection ability of EPIT detection technology for photovoltaic cell defects.Secondly,for the collected infrared thermography sequence of photovoltaic cell defects,two kinds of supervised learning algorithms,Linear Discriminant Analysis（LDA）and Quadratic Discriminant Analysis（QDA）,are used to process thermography sequence,and the time-frequency characteristics of the thermal wave signal at the defect are used to improve the detection ability.The process results are compared with two traditional processing algorithms of Principal Component Analysis（PCA）and Fitting Correlation Coefficient（FCC）.The information entropy,mean square error and signal-to-noise ratio are used to quantitatively analyze the identification ability of the above algorithm for battery defects.Finally,the application of deep learning technology in the automatic detection of infrared image defects of photovoltaic cells is studied,and the experimental verification is carried out by migration learning technology.Firstly,the infrared image data sets of different types of defects in photovoltaic cells are established and expanded.Then,the three networks of AlexNet,VGG16 and VGG19 are fine-tuned,and the established infrared image data set is trained by transfer learning technology.The experimental results show that AlexNet,VGG16 and VGG19 can effectively classify the infrared image defects of photovoltaic cells,and the accuracy is 96.4%,96.7%and 98.4%,respectively.Based on IRT technology and CNN model,it has shown high application potential in defect detection and automatic identification of silicon-based photovoltaic cells,and can provide a reliable method for research and development（R＆D）,testing,manufacturing,service and maintenance of silicon-based photovoltaic cells. |
PDF Full Text Request