In this article, we propose a deep learning based semantic segmentation model that identifies and segments defects in electroluminescence (EL) images of silicon photovoltaic (PV) cells. The proposed model can differentiate between cracks, contact interruptions, cell interconnect failures, and contact corrosion for both multicrystalline and monocrystalline silicon cells. Our model utilizes a segmentation Deeplabv3 model with a ResNet-50 backbone. It was trained on 17,064 EL images including 256 physically realistic simulated images of PV cells generated to deal with class imbalance. While performing semantic segmentation for five defect classes, this model achieves a weighted F1-score of 0.95, an unweighted F1-score of 0.69, a pixel-level global accuracy of 95.4%, and a mean intersection over union score of 57.3%. In addition, we introduce the UCF EL Defect dataset, a large-scale dataset consisting of 17,064 EL images, which will be publicly available for use by the PV and computer vision research communities.

A silicon heterostructure photovoltaic system fielded for 10 years has been investigated in detail. The system has shown degradation, but at a rate similar to an average Si system, and still within the module warranty level. The power decline is dominated by a nonlinear V$_{\rm {oc}}$ loss rather than more typical changes in I$_{\rm {sc}}$ or Fill Factor. Modules have been evaluated using multiple techniques including: dark and light I–V measurement, Suns-V$_{\rm {oc}}$, thermal imaging, and quantitative electroluminescence. All techniques indicate that recombination and series resistance in the cells have increased along with a decrease of factor 2 in minority carrier lifetime. Performance changes are fairly uniform across the module, indicating changes occur primarily within the cells .