Research On Solar Cell Defect Detection Method Based On SVD And Clustering

With the continuous increase in energy demand and the continuous reduction of non-renewable resources,finding and making good use of green new energy has become the current energy development direction of various countries.Sunlight has become the focus of new energy utilization because of its easy availability and green pollution-free characteristics.The core components of solar power generation are solar cells.The quality of solar cells is directly related to the efficiency and service life of solar power generation module.Therefore,the defect detection of solar cells is particularly important.At this stage,the defect detection of solar cells is mainly performed by manual visual inspection,which is inefficient and prone to problems such as misdetection and missed detection.So as to improve the efficiency and accuracy of defect detection,it is convenient for the development and large-scale production of enterprises.Automated image inspection technology is needed for defect detection.Therefore,the research on the defect detection technology of solar cells has both theoretical value and practical application value.This paper analyzes the difficulties of solar cell defect detection,and at the same time,aims at the problem of the uneven brightness of the solar cell surface and the influence of grid line texture on defect detection.It is proposed that the background is reconstructed as the basis,and the background difference retains the target defect,and finally passed Clustering is a method for unsupervised classification of solar cell defects.The main work of this paper is as follows:(1)Separate the background and highlight defects.First of all,this article is based on image Singular Value Decomposition(SVD)and background reconstruction to obtain the main background information of solar cells,and then proposes an adaptive threshold method to determine the optimal number of singular values using gray-scale deviation,and completely restore the texture in the image.Finally,the background difference method and the cluster-based adaptive threshold segmentation method are used to reduce the influence of uneven brightness and raster line texture,and realize the detection of defective images.(2)Feature extraction of surface defects of solar cells.In order to extract the characteristics of the surface defects of the solar cells,the features are extracted from the three aspects of geometry,texture and gray characteristics,and the extracted multi-dimensional data including area,compactness,perimeter,duty cycle and entropy are used to represent the solar cells.By analyzing the distribution curve of the extracted features,the distinctive features are selected for subsequent classification processing.(3)Clustering of surface defects of solar cells.Considering that most of the currently trained classifiers are used to classify defects.In order to make the classification results more reasonable and explanatory,this paper proposes to use a clustering method to divide the extracted defects into different classes and clusters according to specific criteria,Which makes the similarity within clusters large and the similarity between clusters small.Without prior knowledge,the defect features are classified based on the similarity between features to realize the classification of solar cell defects.It adds new possibilities for the defect identification method of solar cells.Finally,the experimental analysis is carried out according to the method proposed in this paper.The experimental results show that the method designed in this paper takes into account the detection and classification of defects.Compared with traditional detection methods,it is the completeness has been improved,it is closer to the actual size of the defect,and the robustness of the defect detection method has been appropriately enhanced.Has certain application prospects.