Presentation Details
| Solar Nowcasting with Local Sensor Networks: Replacing Pyranometers with Abundant IoT Light Sensors and PV Modules (yes) Tobias Veihelmann, Philipp Reitz, Maximilian Lübke, Eva Russwurm, Norman Franchi. Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany |
Abstract
This work presents a hybrid sensor network approach for ultra-short-term solar nowcasting by combining PV modules and automotive-grade ambient light sensors. Unlike traditional local sensor networks that rely exclusively on pyranometers or require using PV plants as both sensor and forecasting target, this scalable approach leverages IoT devices without dedicated deployment infrastructure. The network consists of six 30 W PV modules and four Vishay VEML6031X00 light sensors distributed over approximately 1000 m × 500 m, capturing measurements at one-second resolution. A signal processing chain converts power and illuminance measurements to clear-sky indices through transposition and reverse transposition models, enabling LASSO regression forecasting for horizons of 15-60 seconds. Results from 15 days of intermittent cloud conditions demonstrate RMSE skill scores over smart persistence of 2.86% (15 s), 12.54% (30 s), 13.65% (45 s), and 11.65% (60 s), with the 60-second horizon performance approaching that of precision pyranometer networks. This proof-of-concept demonstrates the viability of cost-effective, scalable sensor networks integrating existing PV systems and IoT devices for practical solar forecasting applications.
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.