Parabolic Trough Solar Field Condition Assessment Using Raw Time Series Data and Artificial Intelligence
- 1Deutsches Zentrum für Luft- und Raumfahrt e. V. (DLR)
Abstract
This work presents a methodology for assessing the condition of parabolic trough solar fields using raw time series data from standard sensors combined with self-supervised time series models. Self-supervised learning enables the extraction of meaningful data patterns without the need for labeled datasets, making it particularly suited for large-scale operational data from solar thermal power plants. Real data from the Andasol 3 parabolic trough field are used to demonstrate the approach. The proposed plant condition model transforms raw sensor data into compact representations that capture essential aspects of solar field operation. A comparison with measured gloss values shows that the learned representations reflect the seasonal evolution of soiling, even though no explicit soiling information was provided to the model. These representations can further serve as universal input for downstream tasks such as anomaly detection, classification, or performance prediction. For example, anomaly detection results demonstrate that events such as soiling episodes or maintenance periods can be identified directly from operational data, despite not being explicitly known to the model.
Related
- Using artificial intelligence for clinical trial quality assurance in a multi-institutional neurovascular-sparing prostate radiotherapy trial
- Future of Artificial Intelligence for Science in Japan 2024 Community Report
- Advancing Nuclear Physics with Machine Learning and Artificial Intelligence
- Artificial intelligence and machine learning for magnetic nanoparticles and nanocarriers in cancer imaging, drug delivery, and magnetic control: a systematic review
- From proof of concept to practice: the next decade of artificial intelligence in metasurface photonics
BibTeX
@article{brenner2026parabolic,
title = {{Parabolic Trough Solar Field Condition Assessment Using Raw Time Series Data and Artificial Intelligence}},
author = {Alex Brenner and Tim Kotzab and Tobias Hirsch},
journal = {SolarPACES Conference Proceedings},
year = {2026},
month = oct,
doi = {10.52825/solarpaces.v4i.2966},
url = {https://doi.org/10.52825/solarpaces.v4i.2966},
note = {CC BY},
}
RIS
TY - JOUR
TI - Parabolic Trough Solar Field Condition Assessment Using Raw Time Series Data and Artificial Intelligence
AU - Alex Brenner
AU - Tim Kotzab
AU - Tobias Hirsch
PY - 2026
DA - 2026/10/23
JO - SolarPACES Conference Proceedings
DO - 10.52825/solarpaces.v4i.2966
UR - https://doi.org/10.52825/solarpaces.v4i.2966
AB - This work presents a methodology for assessing the condition of parabolic trough solar fields using raw time series data from standard sensors combined with self-supervised time series models. Self-supervised learning enables the extraction of meaningful data patterns without the need for labeled datasets, making it particularly suited for large-scale operational data from solar thermal power plants. Real data from the Andasol 3 parabolic trough field are used to demonstrate the approach. The proposed plant condition model transforms raw sensor data into compact representations that capture essential aspects of solar field operation. A comparison with measured gloss values shows that the learned representations reflect the seasonal evolution of soiling, even though no explicit soiling information was provided to the model. These representations can further serve as universal input for downstream tasks such as anomaly detection, classification, or performance prediction. For example, anomaly detection results demonstrate that events such as soiling episodes or maintenance periods can be identified directly from operational data, despite not being explicitly known to the model.
SN - 2751-9899
ER -