Research And Realization Of Solar Cell Flake Detection Method

As the concept of environmental protection has become more and more popular,countries are increasingly reducing the amount of one-time energy exploitation at the expense of the ecological environment,they are turning their attention to cleaner energy,solar energy.As solar cells are the basic carrier of solar photoelectric conversion,the quality of solar cells determines the efficiency of the entire light energy generation.How to "put the last one off’ before the solar cell was shipped out of the factory,the battery could be graded by the quality,which has important theoretical research significance and actual economic value.It solves the problem of detecting the defect area under the flower category(including front scratches,spots,watermarks,cleaning,and fingerprints and so on)and the color difference.This paper presents the corresponding detection algorithm and completes the design of the complete detection system.The main work as the following:In the process,this paper proposes an image correction method based on bilinear interpolation for solar cells.Using the unique texture feature of the cell the positions of the key control points on the cell are obtained by pixel projection normalization,and the distortion image is corrected by the bilinear interpolation algorithm based on template matching.The problem of nonlinearly distorted caused by the angle of the phase and the skew of the edge of the board surface during the phase taking of the battery sheet,etc.was solved.In the detection of defect areas,this paper proposes an Ada-Boost defect area recognition algorithm based on Haar features.First,the cell grid line filling algorithm is used to overcome the interference of the grid texture in the subsequent processing.Then,the Haar-like feature description operator is used to extract the image texture features,and the Ada-Boost cascade classifier is used as input to detect the defect areas Model training.It solves the problem of defect area recognition under the flower category,and its correct rate is 97%.In the process of color difference analysis,this paper proposes a color space clustering method to detect color difference.The color center cluster is obtained by color K-means clustering in the LAB color space,and the degree of chromatic aberration is represented by measuring the color distance.The problem that it is difficult to distinguish the color difference of the color distribution of the battery piece is solved,and it makes up for the blank of the color difference judgment design under the flower piece category.At the stage of the software and hardware system construction under the flower category,the design is completed practical engineering tasks such as hardware selection,software writing,background algorithm import,interface design,etc.,and after a large number of sample training and testing,it still work well.In summary,the flower type detection system designed in this paper can not only solve the problems existing in the current flower detection but also reduce the cost of manual quality inspection to further promote the development of the solar photovoltaic industry.