An evidence-based likelihood approach for the reliability of a complex system with overlapped failure data

verfasst von: Lechang Yang
Abstract: The reliability evaluation of a complex mechanical system is imperative yet challenging because the experiment of a full-scale system is usually unavailable or prohibitively expensive leading to insufficient or incomplete data. Moreover, the collected data is essentially dependent since it is collected from the same system within the same time period, leading to the so-called “overlapped” failure data. To address the dependence between overlapped data in system reliability analysis, a novel concept called Evidence Likelihood Function (ELF) is developed to decompose the original joint likelihood function. This approach is capable of incorporating dependent evidence in the Bayesian framework and provides us with a better understanding of the nature of dependent evidence in system reliability analysis. It has the potential to optimize the system configuration using less full-scale test data in terms of reliability improvement with lower experiment cost.
Organisationseinheit(en): Institut für Risiko und Zuverlässigkeit
Externe Organisation(en): University of Science and Technology Beijing
City University of Hong Kong
Typ: Artikel
Journal: Computers and Industrial Engineering
Band: 201
Anzahl der Seiten: 8
ISSN: 0360-8352
Publikationsdatum: 21.01.2025
Publikationsstatus: Elektronisch veröffentlicht (E-Pub)
Peer-reviewed: Ja
ASJC Scopus Sachgebiete: Allgemeine Computerwissenschaft, Allgemeiner Maschinenbau
Elektronische Version(en): https://doi.org/10.1016/j.cie.2025.110893 (Zugang: Geschlossen)

BibTeX

@article{77afe60e8e034e5e9e29169f9ac15a2c,
title = "An evidence-based likelihood approach for the reliability of a complex system with overlapped failure data",
abstract = "The reliability evaluation of a complex mechanical system is imperative yet challenging because the experiment of a full-scale system is usually unavailable or prohibitively expensive leading to insufficient or incomplete data. Moreover, the collected data is essentially dependent since it is collected from the same system within the same time period, leading to the so-called “overlapped” failure data. To address the dependence between overlapped data in system reliability analysis, a novel concept called Evidence Likelihood Function (ELF) is developed to decompose the original joint likelihood function. This approach is capable of incorporating dependent evidence in the Bayesian framework and provides us with a better understanding of the nature of dependent evidence in system reliability analysis. It has the potential to optimize the system configuration using less full-scale test data in terms of reliability improvement with lower experiment cost.",
keywords = "Bayesian method, Dependent data, Likelihood function, Probability density function, System reliability",
author = "Lechang Yang",
note = "Publisher Copyright: {\textcopyright} 2025 Elsevier Ltd",
year = "2025",
month = jan,
day = "21",
doi = "10.1016/j.cie.2025.110893",
language = "English",
volume = "201",
journal = "Computers and Industrial Engineering",
issn = "0360-8352",
publisher = "Elsevier Ltd.",
}