Detection of Defects in Photovoltaic Modules Through Electroluminescence Image Analysis

Alvim, Amanda Anaice, Figueiredo, Gilberto, Pinho, André da Costa

ISES Solar World Congress 2025 · Fortaleza, Brazil · 2025-11-03
Published by International Solar Energy Society (ISES)
DOI: 10.18086/swc.2025.03.18

Abstract

This work explores the use of machine learning techniques to automate defect detection in photovoltaic (PV) modules through electroluminescence (EL) image analysis. Four approaches were examined: Convolutional Neural Networks (CNNs), Wavelet Scattering with traditional classifiers, Wavelet Scattering combined with CNNs, and a modified VGG16 architecture. Experimental results, based on a dataset of 2,435 EL images from monoand polycrystalline silicon PV cells, showed high classification accuracy—especially when preprocessing techniques were applied. These findings support the integration of EL imaging and machine learning for fault diagnosis in PV systems.

Keywords

Electroluminescence, solar cell defects, machine learning, wavelet scattering

← All proceedings papers