Skip to Main content Skip to Navigation

Modelling and characterization of extreme fatigue risks for metallic materials

Abstract : This work is motivated by a series of questions raised by an industrial issue in material reliability and is divided into two parts. The first one consists in estimating an extreme failure quantile from trials whose outcomes are reduced to indicators failure of the tested specimen. Making use of a splitting approach, we propose a sequential design method which gradually targets the tail of the distribution by sampling under truncated distributions. The model is GEV or Weibull, and is estimated through an improved maximum likelihood procedure for binary data.The second axis aims at developing methodological tools to model fatigue life. To this end, we propose a first test method on composite hypotheses for data affected by additive noise. We handle the problem of maximal decrease of the power for tests on this kind of corrupted data. Comparisons of such tests are considered based on their performances with respect to the Neyman Pearson test between least favourable hypotheses. The second test procedure aims at testing for the number of components of a mixture distribution in a parametric setting. The test statistic is based on divergence estimators derived through the dual form of the divergence in parametric models. We provide a standard limit distribution for the test statistic under the null hypothesis, that holds for mixtures of any number of components k>2.
Complete list of metadata
Contributor : Emilie Miranda Connect in order to contact the contributor
Submitted on : Tuesday, March 1, 2022 - 5:23:06 PM
Last modification on : Friday, August 5, 2022 - 3:00:05 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03264959, version 3


Emilie Miranda. Modelling and characterization of extreme fatigue risks for metallic materials. Statistics [math.ST]. Sorbonne Université, 2020. English. ⟨NNT : 2020SORUS368⟩. ⟨tel-03264959v3⟩



Record views


Files downloads