Research on cable condition prediction method driven by operation data and environmental information
AIP Advances · 01.11.2026 · 16 (11)
DOI: 10.1063/5.0361713 ↗Kurz gefasst
Fachartikel in der Zeitschrift AIP Advances, online erschienen am 12.11.2026. Verfasst von Gang Wang und 4 weiteren Autor:innen. Die Arbeit nennt 20 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: State Grid Shanxi Electric Power Co., Ltd. Science and Technology Project: Research and Application of High-Voltage Cable State Evaluation and Assistant Decisi….
Abstract (Originalsprache)
High-voltage cable insulation aging is a slowly evolving, nonlinear process jointly driven by electrical stress, load fluctuations, and meteorological conditions, and its forward-looking prediction remains a major challenge for conventional algorithms, which typically exhibit pronounced hysteresis, weak environmental generalization, and unsatisfactory long-horizon accuracy. This paper proposes a data- and environment-driven hybrid time-series framework for high-voltage cable health evolution. The framework directly ingests historical health-index sequences, load records, and meteorological parameters as multi-source asynchronous inputs, and an exclusive sliding-window timing feature extraction module derives trend terms, periodic components, and load-environment coupling features from the raw series. A Time-Large Language Model (LLM)-Bidirectional Long Short-Term Memory (BiLSTM) hybrid model then generates multi-scale health trend curves over 7-, 15-, and 30-day horizons and triggers graded advance fault warnings, with the time-LLM capturing multi-year long-term periodic aging rules of cable insulation and BiLSTM tracking short-term dynamic fluctuations of the operating state. Guided by the Arrhenius thermal-aging equation, a transformer-convolutional neural network remaining useful life quantification model translates the predicted trajectories into life estimates, and an online incremental dynamic calibration mechanism continuously adapts the pre-learned mapping to varying load and meteorological working conditions. Independent rolling time-series simulation on three years of real field data demonstrates that the 7-day trend fitting coefficient reaches 0.76, the effective warning rate of hidden faults reaches 82.1%, and the warning false-alarm rate is controlled below 10%. The proposed framework thereby realizes forward-looking trend deduction and life quantification for high-voltage cables, providing quantitative support for predictive maintenance planning.
Förderung
- State Grid Shanxi Electric Power Co., Ltd. Science and Technology Project: Research and Application of High-Voltage Cable State Evaluation and Assistant Decisi…
Zitieren
APA
Wang, G., Guo, D., Shen, J., Zhao, Y., & Guo, R. (2026). Research on cable condition prediction method driven by operation data and environmental information. AIP Advances, 16(11). https://doi.org/10.1063/5.0361713
Vancouver
Wang G, Guo D, Shen J, Zhao Y, Guo R. Research on cable condition prediction method driven by operation data and environmental information. AIP Advances. 2026;16(11). doi:10.1063/5.0361713
BibTeX
@article{Wang2026,
author = {Wang, Gang and Guo, Da and Shen, Jiaming and Zhao, Yitong and Guo, Rui},
title = {Research on cable condition prediction method driven by operation data and environmental information},
journal = {AIP Advances},
year = {2026},
volume = {16},
number = {11},
doi = {10.1063/5.0361713}
}
EUVIMED