Key Technologies Research On The Visual Detection For Automatic Welding Of Solar Cells

With the serious situation of the increasing depletion of the traditional energy, serious issueof environment pollution, and growing demand of energy by the rapid development of theworld economy, the development and utilization of solar energy, a clean renewable energy, hascaused extensive concern from the world and has been developing rapidly. There are many linksto assure the quality of solar cell modules, the main carriers of the development and utilizationof solar energy. The welding of solar cells is the crucial link. Detection of solar cells beforewelding is essential to assure the quality. This paper mainly studies the key technologies on thevisual detection for automatic welding of solar cells, achieving automatic identification forcommon appearance defects of solar cells and precise position measurement of qualified solarcells, which lays a good foundation for automatic welding of solar cells.The main contents of this paper are the followings:(1)Summarize the automatic welding process of solar cells, build hardware system of thevisual detection according to the real condition and introduce system composition and workingprinciple.(2)Analyze the specifications and the classification standard of solar cells, study thepreprocessing techniques of the image, and determine the preprocessing method according tothe characteristics of the image.(3)Analyze the characteristics of the edge breakage defects and use a preliminaryappearance detection method by analyzing contour which is extracted by the boundary tracking.Then complete the edge breakage detection and deflection angle measurement through cornerdetection of contour and the least square method.(4)Complete the broken electrodes detection by analyzing grayscale difference throughscanning the center position of electrodes and complete the crack detection by clearing all theelectrodes, which is based on the image rotation correction. Acquire the image center positionby the coordinate transformation of the image rotation.(5)Design the process of software system in detail and develop software program byLabVIEW to experiment the method which is proposed in this paper. The results show that it isfeasible and effective.