A novel method for detecting low-energy front glass cracks in photovoltaic modules using daylight electroluminescence imaging

This paper proposes a novel application of daylight electroluminescence (EL) imaging for revealing low-energy glass cracks in photovoltaic (PV) modules. These cracks are typically difficult to detect from drone visual RGB or thermography images and require closeup inspection. Stationary EL records only silicon emission and therefore fails to show glass cracks. However, when daylight EL is captured in motion, whether from a drone, another non-stable platform or by manually inducing motion in a fixed setup, crack patterns appear due to angle-dependent reflections from fractured glass; as successive frames are acquired at slightly different camera-to-module angles, these features accumulate in the reconstructed image. The technique is therefore inherently limited to inspections with illumination. Using module selection, tracking, and Fast Fourier Transform (FFT)-based signal extraction, we demonstrate clear crack visibility under daylight while retaining conventional EL defect information. With a 640 × 512 Indium Gallium Arsenide (InGaAs) camera, optimal performance was observed at 8–12 m camera distance, with degraded reliability beyond 15 m. Detection is restricted to the illuminated and imaged glass side, and crack appearance depends on illumination/viewing geometry. A daylight, drone-based EL inspection conducted at the University’s PV plant successfully identified two modules exhibiting glass cracks that were undetectable in infrared thermography or RGB imagery.