Skip to Main content Skip to Navigation

Planification des essais accélérés : optimisation, robustesse et analyse

Abstract : Qualification of a product during the development phases is an important step in a project. It permits to verify that the performances and reliability achieve the objectives. The qualification tests are often costly in time and number of products tested. Accelerated tests consist of submitting units to stress levels higher than operating condition to reduce the time to failures and allow building faster the reliability function. This requires the determination or development of appropriate life-stress relationship model. In addition, a test plan should be built, accurate the design parameters (number of test levels, stress levels, sample allocation at each stress level ...) to find the best compromise between cost test and quality estimation. The objective of the thesis is to define a methodology to obtain optimal and robust accelerated test plans. So, we have developed a general framework based on the minimization of a global cost function and a Bayesian approach. The prior distributions, from the available knowledge on the parameters of reliability and acceleration model, are used in Bayesian inference and in a Monte Carlo simulation to explore possible reliabilities. The optimal and robust plan is obtained by optimization methods (Response surface and Genetic algorithms). Finally, a methodology of monitoring during test realization is developed. The relevance of the results in relation to prior information is studied using a similarity factor between observed and a priori data. It allows verifying whether the decision on the qualification can be taken more quickly in presence of more reliable product than expected or optimizing the plan under realization.
Document type :
Complete list of metadata

Cited literature [84 references]  Display  Hide  Download
Contributor : Anne-Marie Plé Connect in order to contact the contributor
Submitted on : Wednesday, June 11, 2014 - 11:45:43 AM
Last modification on : Wednesday, October 20, 2021 - 3:19:19 AM
Long-term archiving on: : Thursday, September 11, 2014 - 11:35:49 AM


  • HAL Id : tel-01004379, version 1


Seyyedeh Zohreh Fatemi. Planification des essais accélérés : optimisation, robustesse et analyse. Autre. Université d'Angers, 2012. Français. ⟨tel-01004379⟩



Les métriques sont temporairement indisponibles