EUVIMEDEUVIMED entdecken
Crossref

Environmental Data Reliability for Solar Energy Prediction with Deep Learning-Based Anomaly Detection

C. Zargane, A. Kabbori, A. Guezzaz, S. Benkirane

Mathematical Modeling and Computing · 2026 · 13 (2) · 531-543

DOI: 10.23939/mmc2026.02.531 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Mathematical Modeling and Computing, 2026. Verfasst von C. Zargane und 3 weiteren Autor:innen. Die Arbeit nennt 22 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Recently, it is crucial to identify anomalies on environmental data owing to the fact that abnormal patterns can signify a malfunctioning sensor, an error in the system, or a change in the environmental surroundings. Optimization, prevention of system failure, and improvement of data quality require attention to abnormal behavior such as sudden shifts in solar radiation or temperature. Performing such functions can be very challenging with traditional methods, but deep learning approaches have proven to be a great solution in these cases as they are capable of identifying features in large, complex, multidimensional datasets without prior selection of features of interest. In this paper, we used three deep learning models, Autoencoder, Long Short-Term Memory (LSTM), and a hybrid model that combines Convolutional Neural Network (CNN) and LSTM to detect anomalies in environmental datasets. Specifically, we inject synthetic anomalies into temperature at 2 m (T2M) and relative humidity at 2 m (RH2M), representing 10% of the dataset, to assess model performance. Results indicate that the LSTM model outperformed others, achieving 81.16% accuracy for normal data and 83.95% for anomalous data. The Autoencoder followed with 78.51% and 80.16%, respectively. However, the CNN-LSTM model performed worst with detection rates of 69.02% and 71.82%. These results confirm the benefits of deep learning methodologies, and particularly LSTM, in the further development of the reliability of environmental data for solar energy prediction.

Zitieren

APA

Zargane, C., Kabbori, A., Guezzaz, A., & Benkirane, S. (2026). Environmental Data Reliability for Solar Energy Prediction with Deep Learning-Based Anomaly Detection. Mathematical Modeling and Computing, 13(2), 531-543. https://doi.org/10.23939/mmc2026.02.531

Vancouver

Zargane C, Kabbori A, Guezzaz A, Benkirane S. Environmental Data Reliability for Solar Energy Prediction with Deep Learning-Based Anomaly Detection. Mathematical Modeling and Computing. 2026;13(2):531-543. doi:10.23939/mmc2026.02.531

BibTeX
@article{Zargane2026,
  author = {Zargane, C. and Kabbori, A. and Guezzaz, A. and Benkirane, S.},
  title = {Environmental Data Reliability for Solar Energy Prediction with Deep Learning-Based Anomaly Detection},
  journal = {Mathematical Modeling and Computing},
  year = {2026},
  volume = {13},
  number = {2},
  pages = {531--543},
  doi = {10.23939/mmc2026.02.531}
}
Kostenlos starten →