Szczegóły publikacji
Opis bibliograficzny
Design and validation of low-cost MEMS sensor networks for AI-based predictive maintenance in Industry 4.0 / Julia JELEŃSKA, Urszula JACHYMCZYK, Paweł KNAP, Maciej Wiszniewski // W: IEEE MetroXRAINE 2025 [Dokument elektroniczny] : 2025 IEEE International Conference on Metrology for eXtended Reality, Artificial Intelligence and Neural Engineering : 22-24 October 2025, Ancona, Italy : conference proceedings. — Wersja do Windows. — Dane tekstowe. — Piscataway : IEEE, cop. 2025. — Dod. ISBN: 979-8-3315-0280-5 (print on demand). — e-ISBN: 979-8-3315-0279-9. — S. 866-870. — Wymagania systemowe: Adobe Reader. — Bibliogr. s. 870, Abstr.
Autorzy (4)
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 170313 |
|---|---|
| Data dodania do BaDAP | 2026-10-30 |
| Tekst źródłowy | URL |
| DOI | 10.1109/MetroXRAINE66377.2025.11340048 |
| Rok publikacji | 2025 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
| Wydawca | Institute of Electrical and Electronics Engineers (IEEE) |
Abstract
The widespread adoption of predictive maintenance (PdM) in Industry 4.0 hinges on affordable, easily deployable vibration sensors capable of scaling from a handful of monitoring points to factory-wide arrays. Conventional piezoelectric accelerometers provide high fidelity but are prohibitively expensive when hundreds of channels are required. This study introduces a cost-effective PdM platform built around triaxial MEMS accelerometers (IIS3DWB) and ESP32 microcontrollers, implemented in two hardware variants: a battery-powered node for autonomous field deployment and a USB-powered node for laboratory evaluation. Both designs sample three-axis vibration at 4 kHz with 16-bit resolution, stream synchronized data via high-speed UART, and support future wireless extensions. A comprehensive measurement campaign on a rotor- bearing test rig compared MEMS nodes with a reference piezoelectric system under healthy and artificially imbalanced conditions. Machine-learning models―Random Forest, Support Vector Machine (SVM), k-Nearest Neighbors, XGBoost and Decision Tree―were trained on MEMS data and evaluated with leave-one-run-out cross-validation. MEMS sensors achieved ≥ 99 % accuracy across all classifiers; the SVM reached 100 % sensitivity and specificity, outperforming the best piezo result (98.6 %). The implicit 4 kHz bandwidth of the MEMS devices suppressed broadband noise, enhancing low-frequency fault signatures relative to wide-band piezo sensors. This research bridges the gap between costly, high-accuracy industrial sensing and affordable, AI-enhanced alternatives, aligning with Industry 4.0 trends to improve operational reliability, reduce maintenance costs, and enhance asset performance.
