PREDICTING MODULE I-V CURVES FROM ELECTROLUMINESCENCE IMAGES WITH DEEP LEARNING

摘要

Electroluminescence images have been used to qualify the performance of PV modules. Yet, to assess the status and techno-economic performance of a module in a string, quantitative information is required. In this work, we propose a method to predict PV module IV curves from electroluminescence images using a deep learning algorithm. The proposed method consists of creating eleven deep learning models that predict ten points on the IV curve, including I_(SC), I_(mpp), V_(mpp) and V_(OC). We test this method on a dataset of 574 electroluminescence images and IV curves with one dominant fault: inactive cell areas. Results show that the deep learning models are able to find a relationship between the inactive areas of the PV module electroluminescence image and the PV module IV curve. For the test dataset, we predict IV curves with good accuracy values and a mean absolute error for module power below 5 W.