Discrimination between accelerated life models via Approximate Bayesian Computation - Université d'Angers Accéder directement au contenu
Article Dans Une Revue Quality and Reliability Engineering International Année : 2023

Discrimination between accelerated life models via Approximate Bayesian Computation

Résumé

Accelerated life testing (ALT) is widely used in high-reliability product estima-tion to get relevant information about an item’s performance and its failuremechanisms. To analyse the observed ALT data, reliability practitioners needto select a suitable accelerated life model based on the nature of the stress andthe physics involved. A statistical model consists of (i) a lifetime distributionthat represents the scatter in product life and (ii) a relationship between lifeand stress. In practice, several accelerated life models could be used for thesame failure mode and the choice of the best model is far from trivial. Forthis reason, an efficient selection procedure to discriminate between a set ofcompeting accelerated life models is of great importance for practitioners. Inthis paper, accelerated life model selection is approached by using the Approx-imate Bayesian Computation (ABC) method and a likelihood-based approachfor comparison purposes. To demonstrate the efficiency of the ABC method incalibrating and selecting accelerated life model, an extensive Monte Carlo simu-lation study is carried out using different distances to measure the discrepancybetween the empirical and simulated times of failure data. Then, the ABC algo-rithm is applied to real accelerated fatigue life data in order to select the mostlikelymodelamongfiveplausiblemodels.IthasbeendemonstratedthattheABCmethod outperforms the likelihood-based approach in terms of reliability pre-dictions mainly at lower percentiles particularly useful in reliability engineeringand risk assessment applications. Moreover, it has shown that ABC could miti-gate the effects of model misspecification through an appropriate choice of thedistance function.
Fichier principal
Vignette du fichier
Quality Reliability Eng - 2023 - Rabhi - Discrimination between accelerated life models via Approximate Bayesian.pdf (1.41 Mo) Télécharger le fichier
Origine : Fichiers éditeurs autorisés sur une archive ouverte

Dates et versions

hal-03993798 , version 1 (17-02-2023)

Licence

Paternité

Identifiants

Citer

Mohamed Rabhi, Anis Ben Abdessalem, Laurent Saintis, Bruno Castanier, Rodrigue Sohoin. Discrimination between accelerated life models via Approximate Bayesian Computation. Quality and Reliability Engineering International, 2023, ⟨10.1002/qre.3283⟩. ⟨hal-03993798⟩
31 Consultations
28 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More