Contribution to the design of a hybrid machine learning-based predictive maintenance method: The case of the shop process in the biopharmaceutical industry
Design+ · 20.11.2026 · 026300035
DOI: 10.36922/dp026300035 ↗Kurz gefasst
Fachartikel in der Zeitschrift Design+, online erschienen am 20.11.2026. Verfasst von Florent Wadel und 3 weiteren Autor:innen.
Abstract (Originalsprache)
In the biopharmaceutical industry, pneumatic valve diaphragms are critical components whose progressive degradation can affect production reliability and maintenance planning. This paper lays the foundation for a hybrid predictive maintenance approach for valve diaphragms used in the biopharmaceutical industry, within the framework of Prognostics and Health Management. The proposed data-driven method combines supervised and unsupervised machine learning techniques to classify equipment degradation stages and estimate its remaining useful life (RUL). The data, enriched through dedicated feature transformations, enable the identification of the equipment life stages and the prediction of the time remaining before failure. The present work focuses on the design and evaluation of these prognostic models. The integration of the resulting predictions into a decision-support system comparing time-based preventive maintenance with predictive strategies is identified as future work. Although developed for valve diaphragms, the approach is adaptable to other industrial equipment and aims to enhance system reliability while reducing production downtime, in compliance with the regulatory constraints of the biopharmaceutical industry.
Zitieren
APA
Wadel, F., Tighazoui, A., Houssin, R., & Coulibaly, A. (2026). Contribution to the design of a hybrid machine learning-based predictive maintenance method: The case of the shop process in the biopharmaceutical industry. Design+, 0(0), 026300035. https://doi.org/10.36922/dp026300035
Vancouver
Wadel F, Tighazoui A, Houssin R, Coulibaly A. Contribution to the design of a hybrid machine learning-based predictive maintenance method: The case of the shop process in the biopharmaceutical industry. Design+. 2026;0(0):026300035. doi:10.36922/dp026300035
BibTeX
@article{Wadel2026,
author = {Wadel, Florent and Tighazoui, Ayoub and Houssin, Rémy and Coulibaly, Amadou},
title = {Contribution to the design of a hybrid machine learning-based predictive maintenance method: The case of the shop process in the biopharmaceutical industry},
journal = {Design+},
year = {2026},
volume = {0},
number = {0},
pages = {026300035},
doi = {10.36922/dp026300035}
}
EUVIMED