System failure estimation based on field data and semi-parametric modeling. (July 2019)

• Record Type: Journal Article
• Authors: Vališ, David; Pokora, Ondřej; Koláček, Jan
• Is Part Of: Engineering failure analysis. Volume 101(2019)
• Page Start: 473
• Page End: 484
• Publication Date: 2019-07
• Subjects: Oil field data -- Functional data analysis -- Generalized additive models -- Ornstein-Uhlenbeck process -- First hitting time -- Residual useful life; System failures (Engineering) -- Periodicals; Fracture mechanics -- Periodicals; Reliability (Engineering) -- Periodicals; Pannes -- Périodiques; Rupture, Mécanique de la -- Périodiques; Fiabilité -- Périodiques; Fracture mechanics; Reliability (Engineering); System failures (Engineering); Periodicals; Electronic journals; 620.112
• DOI: 10.1016/j.engfailanal.2019.04.014
• Languages: English
• ISSNs: 1350-6307

Abstract

Abstract: A top-priority task nowadays is to ensure quality, safety, and dependability of technical systems. As present systems are highly reliable, it is relatively unlikely for hard failure to occur frequently. One of the ways to avoid failures is by monitoring the conditions and degradation of the system using diagnostic signals. In this article, modern and nontrivial semiparametric approaches to analyze the statistically relevant set of field data are used. In particular, the generalized additive models (GAM) are applied. GAM reflect the current trends in statistics as they include both linear and spline-based modeling. We applied GAM to successfully obtain an appropriate description of the variability of the analyzed field data. The analyzed data come as diagnostic signals from an observed vehicle fleet. Based on the diagnostic signals and applied GAM, we present outcomes from investigating, studying and modeling the technical condition, degradation and failure occurrence of the observed system. Highlights: Contribution to system degradation and failure occurrence understanding based on diagnostics signals. New modeling approach based on functional data analysis. New application of general additive models. Time dependent Ornstein-Uhlenbeck Process utilization. Outcomes applicable as contribution to maintenance optimisation and operation rationalization.