EUVIMEDEUVIMED entdecken
Crossref

Transfer learning-based prediction of aero engine exhaust gas temperature margin using pretrained convolutional neural networks

Xiaonan Chen, Zhiwei Tong, Yuan Liu, Yishou Wang, Xinlin Qing

Engineering Research Express · 31.12.2025 · 7 (4) · 0452i1

DOI: 10.1088/2631-8695/ae2bbf ↗

Kurz gefasst

Fachartikel in der Zeitschrift Engineering Research Express, online erschienen am 29.12.2025. Verfasst von Xiaonan Chen und 4 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 31 Literaturangaben. Gefördert von: Aero Engine Corporation of China (AECC) Hunan Aviation Powerplant Research Institute, National Natural Science Foundation of China, Xiamen City Key Science and Technology Program Project.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Precise performance prediction enhances aero-engine diagnosis accuracy, boosts reliability, and optimizes operational efficiency. Variations between individual engines significantly challenge the performance prediction accuracy. In this study, deep learning and transfer learning were integrated to develop a predictive model architecture for the exhaust gas temperature margin (EGTM) of turbofan engines. Initially, An EGTM prediction model based on double-layer convolutional neural networks (DL-CNN) were constructed, analyzing how different model inputs, prediction intervals, and moving average techniques impacted performance. Building upon this foundation, the transfer learning strategy based on pre-training was employed to enhance EGTM prognostic performance, specifically addressing model accuracy instability due to engine-specific variations. The model validation leverages data from two commercial engines on the same aircraft. The model was trained using data from the left engine and validated with data from the right engine. Under small-sample conditions, transfer learning achieved a reduction in prediction errors of 50%-65%, which demonstrates its effectiveness in mitigating domain shift, addressing data scarcity, and substantially enhancing EGTM prediction accuracy.

Förderung

  • Aero Engine Corporation of China (AECC) Hunan Aviation Powerplant Research Institute
  • National Natural Science Foundation of China
  • Xiamen City Key Science and Technology Program Project

Zitieren

APA

Chen, X., Tong, Z., Liu, Y., Wang, Y., & Qing, X. (2025). Transfer learning-based prediction of aero engine exhaust gas temperature margin using pretrained convolutional neural networks. Engineering Research Express, 7(4), 0452i1. https://doi.org/10.1088/2631-8695/ae2bbf

Vancouver

Chen X, Tong Z, Liu Y, Wang Y, Qing X. Transfer learning-based prediction of aero engine exhaust gas temperature margin using pretrained convolutional neural networks. Engineering Research Express. 2025;7(4):0452i1. doi:10.1088/2631-8695/ae2bbf

BibTeX
@article{Chen2025,
  author = {Chen, Xiaonan and Tong, Zhiwei and Liu, Yuan and Wang, Yishou and Qing, Xinlin},
  title = {Transfer learning-based prediction of aero engine exhaust gas temperature margin using pretrained convolutional neural networks},
  journal = {Engineering Research Express},
  year = {2025},
  volume = {7},
  number = {4},
  pages = {0452i1},
  doi = {10.1088/2631-8695/ae2bbf}
}
Kostenlos starten →