Vanilla LSTM Predictive Maintenance Model for Scientific Research Facilities
Sensors · 20.07.2026 · 26 (14) · 4581
DOI: 10.3390/s26144581 ↗Kurz gefasst
Fachartikel in der Zeitschrift Sensors, online erschienen am 20.07.2026. Verfasst von Edward Nkadimeng und 5 weiteren Autor:innen. Beteiligte Einrichtung: School of Physics and Institute for Collider Particle Physics, University of the Witwatersrand, Johannesburg 2050, South Africa. Hauptthemen laut MEDLINE: Research. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 26 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Carnegie Corporation of New York, National Research Foundation of South Africa, University of the Witwatersrand.
Abstract (Originalsprache)
Ensuring the reliability and operational efficiency of critical scientific equipment is a central challenge in high-stakes research environments such as nuclear physics laboratories and particle accelerator facilities. Unexpected failures entail significant financial cost and prolonged interruptions to experimental programmes. We present a predictive maintenance (PdM) framework built around a two-layer Vanilla Long Short-Term Memory (LSTM) network trained on multivariate sensor streams collected at NRF-iThemba LABS between January 2021 and December 2023. Four channels, namely supply voltage, vibration velocity, differential pressure, and rotational speed, were recorded at 5 min intervals using a suite of industrial-grade transducers (power quality analyser, IEPE accelerometers, differential pressure transmitters, and proximity encoders) feeding a multi-channel data-acquisition chassis via OPC-UA, yielding a time-synchronised dataset of 315,360 observations. A normalised failure score converts the binary classifier output into a continuous, interpretable health indicator that supports tiered scheduling of maintenance. The Vanilla LSTM achieved a test-set F1-score of 75% and an area under the receiver-operating-characteristic curve (AUC) of 0.856, outperforming five competing architectures (PCA/T2, Random Forest, Deep Neural Network, LSTM Autoencoder, and Bidirectional LSTM Autoencoder), and delivered a mean failure lead time of (42.3±7.2)h, exceeding the 36 h engineering requirement for proactive maintenance scheduling.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- School of Physics and Institute for Collider Particle Physics, University of the Witwatersrand, Johannesburg 2050, South Africa.
- Department of Electrical and Electronic Engineering Science, Faculty of Engineering and the Built Environment, University of Johannesburg, Auckland Park, Johannesburg 2006, South Africa.
- iThemba LABS, National Research Foundation, Old Faure Road, Somerset West 7129, South Africa.
ORCID-Profile: Edward Nkadimeng, Manal Karmoude, Mukesh Kumar, Bruce Mellado
Förderung
- Carnegie Corporation of New York
- National Research Foundation of South Africa
- University of the Witwatersrand
Zitieren
APA
Nkadimeng, E., Gololo, M., Karmoude, M., Stodart, N., Kumar, M., & Mellado, B. (2026). Vanilla LSTM Predictive Maintenance Model for Scientific Research Facilities. Sensors, 26(14), 4581. https://doi.org/10.3390/s26144581
Vancouver
Nkadimeng E, Gololo M, Karmoude M, Stodart N, Kumar M, Mellado B. Vanilla LSTM Predictive Maintenance Model for Scientific Research Facilities. Sensors. 2026;26(14):4581. doi:10.3390/s26144581
BibTeX
@article{Nkadimeng2026,
author = {Nkadimeng, Edward and Gololo, Mpho and Karmoude, Manal and Stodart, Nieldane and Kumar, Mukesh and Mellado, Bruce},
title = {Vanilla LSTM Predictive Maintenance Model for Scientific Research Facilities},
journal = {Sensors},
year = {2026},
volume = {26},
number = {14},
pages = {4581},
doi = {10.3390/s26144581}
}
EUVIMED