Research On Color Classification And Defect Detection System Of Solar Cells

As the primary energy sources in diminishing and the destruction of the ecological environment, many countries have turned to the solar energy because it is inexhaustible and clean. As a carrier of the solar energy,the quality of the solar cell affects the life of its module, as well as the stability and the photoelectric conversion efficiency. They are also the main indicator of the market competitiveness among the solar cell manufacturers. The traditional manual inspection and sorting have been unable to satisfy the needs of the solar market.In this paper, some research on the color of the solar cells and their appearance defects will be done based on machine vision theory and image processing. The color classification and defect detection system of solar cells is also constructed, besides, image recognition software is developed for the color polymorphism,diversity of the defect and complexity of defect images of the solar cell. This system realizes the high accuracy of the color separation and the high recognition rate of the defects such as collapse edge, broken-grid, dirty.Meanwhile, this system can meet the requirements of the real-time online detection. Details are as follows.Through in-depth understanding of the domestic and international situation and development trend of solar cells in color separation and defect detection, this paper puts forward the overall design plan and builds the hardware platform of the system on the basis of learning and researching the theory of machine vision and image processing technology. The hardware platform mainly includes the white ball integral light source, area array CMOS gigabit network camera, lens, black box, a switch power supply, data collector, photoelectric sensor, to ensure that it can obtain the images clearly and deliver to the host quickly in the automatic production line.Based on the understanding of the color standards and separation methods in the manufacturer, the paper uses LabVIEW image processing functions to enhance the clarity of the target details by the pretreatment and the logic color operation of the solar cell image. Then it uses adaptive classification algorithm based on the Euclidean theory to meet the different solar cell classification standards in different production enterprises. All of this can make users establish their own solar cell color standard library.The solar cell appearance defect types and detection methods are studied, above all, the algorithm of collapse edge, broken-grid and dirty is designed. In detail, contour analysis and fitting line is used to detect the chipping; the sub-regional and looking for communication domain can be used to identify breaking-gate;image gray difference algorithm based on line scan is used to detect some defects contain plasma leakage, oil spots, fingerprints and other defects. At the same time, this paper improves the recognition rate of the defects of the solar cells by the optimization of the image processing program and the algorithm.Lastly, this study uses LabVIEW to develop the PC interface and serial communication program. The interface is beautiful, sample and east to operate. After the test in the enterprise production line, running speed and detection accuracy can meet the production needs of the enterprise basically. At present, this system is undergoing a transition and actively to the market.