Presentation Details
| Rapid, non-contact testing for PV structural defects using hyperspectral imaging and machine learning (yes) Aman Raj, Dwarakanath Ravikumar, Saurav Kumar. Arizona State University, Tempe, AZ, USA |
Abstract
Structural defects in front glass, such as micro- and macrocracks, significantly compromise the long-term reliability of PV installations. Because these defects are visually undetectable and propagate rapidly, they often cause power loss and localized hotspots. Techniques such as Electroluminescence (EL) imaging and Infrared thermography (IRT) are primarily sensitive to electrically active defects and therefore cannot detect structural cracks in the glass. Moreover, EL imaging requires dark-room conditions, necessitating module disconnection and transport to laboratory facilities, which limits its practicality for field-based diagnostics. This work presents the first rapid, noninvasive, and non-destructive hyperspectral imaging (HSI) and machine learning (ML)–based framework for simultaneously detecting micro- and macro cracks on the glass surface of PV modules using both visible–near-infrared and short-wave infrared spectral data. These findings represent the first demonstration of HSI as a scalable diagnostic tool for early-stage crack detection in PV modules, with the potential to improve the reliability, safety, and maintenance of photovoltaic systems.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.