A hybrid CNN-BiLSTM model integrating meteorological and topographic factors for transmission line ice thickness prediction
Frontiers in Environmental Science · 24.07.2026 · 14
DOI: 10.3389/fenvs.2026.1851382 ↗Kurz gefasst
Fachartikel in der Zeitschrift Frontiers in Environmental Science, online erschienen am 24.07.2026. Verfasst von Panfeng Li und 3 weiteren Autor:innen. Die Arbeit nennt 26 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Science and Technology Project of State Grid.
Abstract (Originalsprache)
Transmission-line icing can threaten reliable grid operation, making short-term ice-thickness prediction important for disaster prevention and operational planning. This study develops a late-fusion CNN-BiLSTM regression framework for next-hour ice-thickness prediction on the 500 kV Changmiao 5P01 transmission-line corridor in Huzhou, Zhejiang, China. A 24-h historical meteorological sequence is processed by 1D convolutional layers and a BiLSTM without using observations after the prediction boundary; location-specific static topographic descriptors are concatenated only with the resulting boundary representation before linear regression. The study uses 2,348 continuous hourly records. Inputs include temperature, relative humidity, atmospheric pressure, precipitation, and wind components, together with elevation, slope, and aspect; ice thickness is measured in millimeters. Under the same chronological train/test protocol, the proposed configuration produced lower test errors than the selected standalone CNN, LSTM, and BiLSTM baseline configurations. The results support the predictive usefulness of the terrain descriptors for the studied subtropical hilly corridor, but do not establish causal effects or broader geographic generalization.
Förderung
- Science and Technology Project of State Grid
Zitieren
APA
Li, P., Feng, B., Fu, Z., & Zhou, J. (2026). A hybrid CNN-BiLSTM model integrating meteorological and topographic factors for transmission line ice thickness prediction. Frontiers in Environmental Science, 14. https://doi.org/10.3389/fenvs.2026.1851382
Vancouver
Li P, Feng B, Fu Z, Zhou J. A hybrid CNN-BiLSTM model integrating meteorological and topographic factors for transmission line ice thickness prediction. Frontiers in Environmental Science. 2026;14. doi:10.3389/fenvs.2026.1851382
BibTeX
@article{Li2026,
author = {Li, Panfeng and Feng, Bing and Fu, Zhenyang and Zhou, Jian},
title = {A hybrid CNN-BiLSTM model integrating meteorological and topographic factors for transmission line ice thickness prediction},
journal = {Frontiers in Environmental Science},
year = {2026},
volume = {14},
doi = {10.3389/fenvs.2026.1851382}
}
EUVIMED