Presentation Details
| pvcracks: python repository for electroluminescence image processing, current-voltage curve fitting and power loss estimation Norman Jost1, Brandon K.Byford1, Rodrigo d.Prado Santamaria2, Clifford W.Hansen1, Jennifer L.Braid1. 1Sandia National Laboratories, Albuquerque, NM, USA.2Technical University of Denmark, Roskilde, Denmark |
Abstract
The recent open-source pvcracks repository enables rapid, automated EL analysis by integrating deep learning for electroluminescence (EL) image segmentation to produce pixel-level masks of cracks, inactive regions, and busbars; parameterizing the masks using a variational autoencoder (VAE); and clustering images by degradation severity. Beyond image processing and segmentation tools, the pvspice subpackage leverages electrical circuit simulation (SPICE) to fit measured current-voltage (IV) curves accurately, synthesize module level EL/IV data from cell level inputs, and model power loss due to spatially inhomogeneous cracking of the solar cells across modules or entire strings. We showcase these different results and demonstrate how researchers and engineers can use pvcracks, through simple, reproducible examples hosted in the repository, to segment EL images, classify degradation severity, fit IV characteristics, and simulate module performance under realistic defect scenarios. The goal is to offer usable tools that allows to accelerate data processing for reliability assessments of PV modules.
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.
Content Locked. Log into a registered attendee account to access this presentation.