EUVIMEDEUVIMED entdecken
Open Access · CC BY-ND 4.0

Self-Optimizing Industrial Intelligence

Chaitanya Nagabhyru

Street Art & Urban Creativity · 10.11.2025 · 11 (7s) · 120-135

DOI: 10.66838/sauc.6548 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Street Art & Urban Creativity, online erschienen am 10.11.2025. Verfasst von Chaitanya Nagabhyru. Der Volltext ist frei zugänglich (Open Access, CC BY-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Unplanned equipment downtime erodes the productivity of U.S. factories, and the machine-learning models built to anticipate it often decay when equipment is upgraded or operating conditions change. This paper presents a multimodal predictive-maintenance framework that fuses machine sensor streams, equipment alarm logs, technician maintenance notes and production data, monitors each data type for drift, and adapts the model when drift is confirmed. A design layer for retrieval-based work-order assistance with expert validation is described but not evaluated. Because plant data cannot be shared, the framework is evaluated on a transparent synthetic fleet of 160 machines over 900 days, in which a retrofit of 40% of machines, a heavier workload and new production mix are introduced at day 450. Fusing all four data types raises average precision for seven-day failure prediction from 0.32 for sensors alone to 0.46 (late fusion) and 0.47 (early fusion) after the change.....

Zitieren

APA

Nagabhyru, C. (2025). Self-Optimizing Industrial Intelligence. Street Art & Urban Creativity, 11(7s), 120-135. https://doi.org/10.66838/sauc.6548

Vancouver

Nagabhyru C. Self-Optimizing Industrial Intelligence. Street Art & Urban Creativity. 2025;11(7s):120-135. doi:10.66838/sauc.6548

BibTeX
@article{Nagabhyru2025,
  author = {Nagabhyru, Chaitanya},
  title = {Self-Optimizing Industrial Intelligence},
  journal = {Street Art & Urban Creativity},
  year = {2025},
  volume = {11},
  number = {7s},
  pages = {120--135},
  doi = {10.66838/sauc.6548}
}
Kostenlos starten →