Research On Online Color Classification Of Solar Cells

With the progress of science and technology,solar energy has been used more and more widely as a kind of clean and pollution-free renewable energy to promote the rapid development of photovoltaic power generation industry.Solar cells are the most major carriers for photovoltaic power generation,and the solar cell most commonly used is silicon solar cell.Due to the different production processes and materials,there are different colors on silicon solar cells' surface,and even one cell may exist different color,namely color difference.Different color cells and cells with color difference have different photoelectric conversion efficiency.And using cells with different photoelectric conversion efficiency together will reduce the overall solar power generation effect.It's necessary to do color difference detection and color classification after the completion of solar cells production.At present,color difference detection and color classification of solar cells are based on manual detection method,which is heavily depended on the visual detection of worker's visual subjectivity,and the detection speed is slow,the accuracy rate is not high.Therefore,there is an urgent need to find an efficient and rapid method to classify solar cells instead of manual detection method.This paper combines the image processing and machine vision technology to study the online color classification system of solar cells,which are studied from the selection of part hardware devices and image color classification algorithm.Aiming at the hardware selection part,this paper analyzes the selection of industrial camera and lenses and the design of lighting system.The algorithm design part of this study is the process of image processing to obtain the solar cell images.By analyzing the characteristics of the 6 types of solar cells,the images are preprocessed by image preprocessing to obtain the ROI.Then,crystal lattice detection model based on the sum of edge images pixels is established.Then,feature extraction of the non-crystal-lattice type solar cells is carried out.A method for selecting the suitable color features of solar cells are proposed by using various features of histograms.After obtaining color features,according to machine vision theory,a color classification model based on color features using support vector machine is established,and cells with and without color difference are distinguished.Then combining the Euclidean distance with SVM algorithm,the accurate color classifications results are obtained.Through the above research,we can get the classification of six types of cells,including crystallattice,color difference,light blue,ink blue,medium blue and dark blue,and get the ideal results.The results show that the algorithm proposed in this study is suitable for the online classifying of solar cells.