Szczegóły publikacji
Opis bibliograficzny
Fault diagnosis using artificial neural networks trained only on signals from an undamaged machine / Paweł PAWLIK, Konrad Kania, Bartosz Przysucha // W: Advances in Technical Diagnostics II : proceedings of the 7th International Congress on Technical Diagnostics, ICTD 2022 : 14–16 September 2022, Radom, Poland / eds. Andrzej Puchalski, [et al.]. — Cham : Springer Nature Switzerland, cop. 2023. — (Applied Condition Monitoring ; ISSN 2363-698X ; vol. 21). — ISBN: 978-3-031-31718-7; e-ISBN: 978-3-031-31719-4. — S. 166–175. — Bibliogr., Abstr. — Publikacja dostępna online od: 2023-05-21
Autorzy (3)
- AGHPawlik Paweł
- Kania Konrad
- Przysucha Bartosz
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 147065 |
|---|---|
| Data dodania do BaDAP | 2023-06-05 |
| DOI | 10.1007/978-3-031-31719-4_17 |
| Rok publikacji | 2023 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
| Wydawca | Springer |
| Czasopismo/seria | Applied Condition Monitoring |
Abstract
Fault diagnosis of machines operating under variable conditions requires advanced signal analysis methods. Variable conditions of load, temperature or rotational speed influence the values of vibration signals. This paper proposes a diagnosis method based on order analysis and an artificial neural network trained solely on data for a machine operating in a fault-free condition. The order spectrum of the new parameter rDPNS, which does not depend on the machine’s working conditions, was proposed. The obtained order spectrum of this parameter allows the identification of faults by the theory of fault diagnostics. The proposed method has been verified in diagnosing the degradation of a two-stage cylindrical gear. The diagnostic experiment was conducted on a laboratory bench. The signals of vibration acceleration, rotational speed, and current supply to the drive motor were recorded at varying load and temperature. The results of the experiment conducted on the laboratory bench showed the effectiveness of the proposed method.
