EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Artificial Intelligence and Machine Learning in Rotordynamics: A Critical Review of Vibration Modeling, Prediction, Diagnosis, and Control

Fahimeh Mehralian, Fevzi Çakmak Bolat

Actuators · 15.11.2026 · 15 (11) · 605

DOI: 10.3390/act15110605 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Actuators, online erschienen am 15.11.2026. Verfasst von Fahimeh Mehralian und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Artificial intelligence and machine learning increasingly support vibration analysis of rotating machinery, yet diagnostic accuracy alone does not establish reliable response prediction or control of a rotor–bearing system. This critical review connects dynamic modeling, response and stability prediction, fault diagnosis, and vibration control and design through a common engineering framework. Learning methods are classified by their target quantities and by how physical information enters features, training data, equation residuals, network architecture, or hybrid corrections. Published experimental setups and results illustrate response generation, staged diagnosis, interacting faults, and unbalance identification. Hybrid models can improve agreement with measured responses, while physics-guided architectures can constrain localization and parameter estimation; these benefits remain dependent on the measured system and operating conditions. Persistent challenges include inverse-problem identifiability, simulation-to-experiment discrepancy, independent-machine generalization, uncertainty calibration, and closed-loop stability. The review relates these challenges to task-specific metrics, appropriate conventional baselines, and experiments that preserve resonance, phase, fault observability, and actuator constraints. An integrated development lifecycle links model construction, independent evaluation, and application qualification. The central finding is that useful learning in rotordynamics requires a demonstrable improvement in an engineering decision within an experimentally supported operating envelope.

Zitieren

APA

Mehralian, F., & Bolat, F. Ç. (2026). Artificial Intelligence and Machine Learning in Rotordynamics: A Critical Review of Vibration Modeling, Prediction, Diagnosis, and Control. Actuators, 15(11), 605. https://doi.org/10.3390/act15110605

Vancouver

Mehralian F, Bolat FÇ. Artificial Intelligence and Machine Learning in Rotordynamics: A Critical Review of Vibration Modeling, Prediction, Diagnosis, and Control. Actuators. 2026;15(11):605. doi:10.3390/act15110605

BibTeX
@article{Mehralian2026,
  author = {Mehralian, Fahimeh and Bolat, Fevzi Çakmak},
  title = {Artificial Intelligence and Machine Learning in Rotordynamics: A Critical Review of Vibration Modeling, Prediction, Diagnosis, and Control},
  journal = {Actuators},
  year = {2026},
  volume = {15},
  number = {11},
  pages = {605},
  doi = {10.3390/act15110605}
}
Kostenlos starten →