Defect detection on solar cells using mathematical morphology and fuzzy logic techniques

Solar cells or photovoltaic systems have been extensively used to convert renewable solar energy to generate electricity, and the quality of solar cells is crucial in the electricity-generating process. Mechanical defects such as cracks and pinholes affect the quality and productivity of solar cells. Thus, it is necessary to detect these defects and reject the defected ones from solar cells production line. Various inspection methods have been proposed based on contact and non-contact methods. The contact methods are usually destructive due to the contact to product, but non-contact methods implemented low accuracy rate or high hardware installation cost. Therefore, it is needed to develop a robust non-contact solar cell inspection method with low hardware installation cost. In this paper, we proposed a non-contact and nondestructive automated visual inspection system that was able to perform defect detection using image processing and fuzzy logic techniques. The image processing techniques involved thresholding, mathematical morphology, and edge detection operators. In order to assess the proposed system, comprehensive evaluation systems were conducted and presented which was consisted of module and integrated evaluations. For the purpose of identifying and categorizing errors, performance comparisons between the production rule, Mamdani, and Sugeno fuzzy models were done. The results of the experiments revealed that, using the Mamdani fuzzy model, the accuracy rates for identifying individual and group defects were 97.08% and 96%, respectively.