Szczegóły publikacji
Opis bibliograficzny
Bayesian inference for transformer degradation prediction: a comparative analysis with traditional and machine learning models / Anna JAROSZ, Jerzy BARANOWSKI // W: IEEE EUROCON 2025 [Dokument elektroniczny] : 21st international conference on Smart technologies : June 4-6, 2025, Gdynia, Poland : proceedings / eds. Ireneusz Czarnowski, Marek Jasiński. — Wersja do Windows. — Dane tekstowe. — [Piscataway] : Institute of Electrical and Electronics Engineers, cop. 2025. — (International Conference on Computer as a Tool ; ISSN 2837-7990). — Print on Demand (PoD) ISBN: 979-8-3315-0879-1. — e-ISBN: 979-8-3315-0878-4. — S. [1–5]. — Wymagania systemowe: Adobe Reader. — Bibliogr. s. [5], Abstr. — Publikacja dostępna online od: 2025-07-15
Autorzy (2)
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 160633 |
|---|---|
| Data dodania do BaDAP | 2025-07-10 |
| Tekst źródłowy | URL |
| DOI | 10.1109/EUROCON64445.2025.11073497 |
| Rok publikacji | 2025 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
| Wydawca | Institute of Electrical and Electronics Engineers (IEEE) |
| Czasopismo/seria | International Conference on Computer as a Tool |
Abstract
Predicting transformer degradation is essential for ensuring the reliability and efficiency of power systems. This study explores Bayesian inference as a robust alternative to traditional statistical and machine learning approaches for degradation modeling. Unlike conventional methods, Bayesian models explicitly quantify uncertainty, making them well-suited for predictive maintenance. The proposed framework incorporates Markov Chain Monte Carlo (MCMC) techniques to improve parameter estimation and model selection. A systematic evaluation is conducted using predictive accuracy, Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and computational efficiency. Results demonstrate that Bayesian models outperform traditional and machine learning methods, with the MCMC Bayesian model achieving the highest predictive accuracy (92.1%) while providing superior uncertainty quantification. Future research directions include enhancing computational efficiency, integrating Bayesian models with real-time monitoring, and validating their effectiveness across diverse transformer systems. These findings contribute to more reliable predictive maintenance strategies, reducing operational risks and improving power network resilience.
