EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Deep hybrid models for daily natural gas consumption forecasting and complexity measuring

Shengxi Bai, Xiaomei Huang, Min Luo, Junfu Su

Energy Science & Engineering · 2023-02 · 11 (2) · 654-674

DOI: 10.1002/ese3.1352 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Energy Science & Engineering, 2022-12. Verfasst von Shengxi Bai und 3 weiteren Autor:innen. Die Arbeit wurde bisher 12-mal zitiert (Crossref) und nennt 41 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurately forecasting daily natural gas consumption (NGC) is challenging due to its complex time‐varying high‐dimensional features. Therefore, this paper proposes a novel hybrid model for daily NGC forecasting consisting of two stages. In Phase I, the parallel local outlier factor‐isolation forest‐based data preprocessing algorithm is applied to denoise input data and preserve valuable features. In Phase II, the convolutional neural network (CNN)–stacked long short‐term memory (SLSTM) can extract and model spatio‐temporal features and is a hybrid deep learning forecaster. The influencing mechanism of different factors on the daily NGC prediction accuracy has not been mathematically and experimentally revealed. Thus, a hybrid complexity measure model () is developed by combining the variation coefficient analysis, local fluctuation coefficient analysis, and kurtosis analysis. The historical data sets of three representative cities were collected in the case studies, and six advanced models were selected for comparison. The results reveal that high‐quality input data preprocessed in Phase I enables better prediction performance for all models. The CNN‐SLSTM outperforms other algorithms with an average prediction improvement of about 26%–49%. Prediction models perform better in low NGC periods and cities with predominantly industrial natural gas users due to their lower data complexity.

Zitieren

APA

Bai, S., Huang, X., Luo, M., & Su, J. (2023). Deep hybrid models for daily natural gas consumption forecasting and complexity measuring. Energy Science & Engineering, 11(2), 654-674. https://doi.org/10.1002/ese3.1352

Vancouver

Bai S, Huang X, Luo M, Su J. Deep hybrid models for daily natural gas consumption forecasting and complexity measuring. Energy Science & Engineering. 2023;11(2):654-674. doi:10.1002/ese3.1352

BibTeX
@article{Bai2023,
  author = {Bai, Shengxi and Huang, Xiaomei and Luo, Min and Su, Junfu},
  title = {Deep hybrid models for daily natural gas consumption forecasting and complexity measuring},
  journal = {Energy Science & Engineering},
  year = {2023},
  volume = {11},
  number = {2},
  pages = {654--674},
  doi = {10.1002/ese3.1352}
}
Kostenlos starten →