Szczegóły publikacji
Opis bibliograficzny
Noise injection framework for development of noise-robust predictive maintenance solutions / Urszula JACHYMCZYK, Paweł KNAP, Adam Szulęcki // 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. 854-859. — Wymagania systemowe: Adobe Reader. — Bibliogr. s. 859, Abstr. — Publikacja dostępna online od: 2026-01-23
Autorzy (3)
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 165993 |
|---|---|
| Data dodania do BaDAP | 2026-03-10 |
| Tekst źródłowy | URL |
| DOI | 10.1109/MetroXRAINE66377.2025.11340190 |
| Rok publikacji | 2025 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
| Wydawca | Institute of Electrical and Electronics Engineers (IEEE) |
Abstract
In predictive maintenance (PdM) applications, sensor signals are often affected by various forms of noise that can degrade model accuracy and reliability. While many studies use Gaussian noise to simulate environmental disturbances, real-world sensor data is typically impacted by a broader range of noise types, including impulsive, frequency-dependent, and drift-based distortions. This paper presents a modular noise injection framework designed to augment PdM datasets with diverse synthetic and measured noise signals. The framework simulates realistic sensor and system-level disturbances and is evaluated on vibration data from an online shaft monitoring test stand. We assess the impact of noise-based augmentation on both limited and comprehensive datasets using six machine learning models. Results demonstrate that properly designed noise injection can substantially enhance model robustness and performance, particularly in data-constrained settings. Notably, we observed that models trained on raw time-domain data, while offering greater potential for learning rich signal characteristics, also exhibit higher sensitivity to aggressive augmentation. Our findings highlight the importance of carefully tuning noise augmentation strategies to match the measurement context and model architecture.
