Classification of silicon solar cells using Electroluminescence texture analysis

An automated procedure for classification of polycrystalline silicon solar cells with respect to their electrical characteristics is presented in this work. Electrical characteristics of solar cells are a very important issue in the photovoltaic panel production process, as they affect the final product quality. The procedure is composed of two sequential steps: in the first step a vector of features is extracted from the Electroluminescence intensity images of photovoltaic cells, making use of a texture analysis technique named Sum and Difference Histogram. In the second step the classification is carried out through a particular structure of Neural Network and a proper decision rule. The technique is especially suited to be implemented in production line, as it is fast and has a low computational complexity. Moreover, experimental results demonstrate the good performances in terms of successful classification.