EUVIMEDEUVIMED entdecken
Crossref

Multidomain Supervised Time–Frequency Feature Fusion Autoencoder for Health Indicator Extraction in Rotating Machinery

Hanbyeol Park, Jungho Choo, Gawon Lee, Hyerim Bae

IEEE Sensors Journal · 2026-08 · 26 (16) · 24239-24264

DOI: 10.1109/jsen.2026.3702985 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Sensors Journal, 2026-08. Verfasst von Hanbyeol Park und 3 weiteren Autor:innen. Die Arbeit nennt 51 Literaturangaben. Gefördert von: Institute of Information & Communications Technology Planning & Evaluation (IITP)-Innovative Human Resource Development for Local Intellectualization program g….

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Constructing health indicators (HIs) that quantitatively represent the state of a machine is essential for the condition-based maintenance of rotating machinery. Traditional physics-based HI (PHI) construction methods have limited effectiveness; for example, virtual HIs (VHIs) integrating multiple features have limited applicability because of the manual feature extraction process that depends on expert knowledge and a lack of model adaptability. Existing studies focus on single domains (raw signals) or use time–frequency domain spectrograms and fail to adequately extract meaningful frequency band patterns extended along specific axes. These limitations in representation capacity lead to a degraded performance in HI extraction. This article proposes a novel deep-learning-based HI extraction method that uses a supervised time–frequency feature fusion autoencoder (AE) for fusing multidomain features. The proposed model includes spatiotemporal feature extraction from raw signals using a multiscale convolutional neural network (CNN) Transformer block, extraction of temporally resolved frequency features from time–frequency maps using rectangular kernels and cross-attention mechanisms, fusion of latent representations to extract 1-D HI, and simultaneous reconstruction of the input signal from the extracted HI unified encoder–decoder architecture. During training, the model optimizes a composite loss function that comprises a shape constraint function (SCF) and domain-specific reconstruction errors using a dynamic weight-averaging strategy. The proposed method is validated using publicly available rotating machinery datasets and demonstrates superior performances across four evaluation metrics compared with those of existing methods. This study highlights the potential of automated high-performance HI extraction for rotating machinery and offers practical contributions to industrial maintenance applications.

Förderung

  • Institute of Information & Communications Technology Planning & Evaluation (IITP)-Innovative Human Resource Development for Local Intellectualization program g…

Zitieren

APA

Park, H., Choo, J., Lee, G., & Bae, H. (2026). Multidomain Supervised Time–Frequency Feature Fusion Autoencoder for Health Indicator Extraction in Rotating Machinery. IEEE Sensors Journal, 26(16), 24239-24264. https://doi.org/10.1109/jsen.2026.3702985

Vancouver

Park H, Choo J, Lee G, Bae H. Multidomain Supervised Time–Frequency Feature Fusion Autoencoder for Health Indicator Extraction in Rotating Machinery. IEEE Sensors Journal. 2026;26(16):24239-24264. doi:10.1109/jsen.2026.3702985

BibTeX
@article{Park2026,
  author = {Park, Hanbyeol and Choo, Jungho and Lee, Gawon and Bae, Hyerim},
  title = {Multidomain Supervised Time–Frequency Feature Fusion Autoencoder for Health Indicator Extraction in Rotating Machinery},
  journal = {IEEE Sensors Journal},
  year = {2026},
  volume = {26},
  number = {16},
  pages = {24239--24264},
  doi = {10.1109/jsen.2026.3702985}
}
Kostenlos starten →