Data Uncertainty and Important Measures

Abstract : This book focuses on modeling several forms of uncertainty (epistemic and aleatory) and dealing with parameter uncertainty in dependability analysis. For this purpose, we are modeling these forms of uncertainty through additive and non-additive theories. Several theories or modeling languages are used. After presenting some context of uncertainty sources and several theoretical frameworks for modeling the different forms of uncertainty, they are applied on assessing the performance of system reliability or dependability with usual dependability models. Industrial systems or toys systems for the sake of illustration are used. Beyond the usual models in dependability, the concept of evidential networks is introduced. Similar to Bayesian networks but considering non additive theories, the modeling principle is explained and applied to several forms of uncertainty and on several systems. This modeling tool is also used to computed importance measures which are necessary to improve systems or test the robustness of the assessment even in the context of several parameter uncertainties.
Document type :
Books
Complete list of metadatas

https://hal.archives-ouvertes.fr/hal-01683009
Contributor : Philippe Weber <>
Submitted on : Friday, January 12, 2018 - 4:36:50 PM
Last modification on : Tuesday, January 8, 2019 - 4:55:05 PM

Identifiers

  • HAL Id : hal-01683009, version 1

Citation

Christophe Simon, Philippe Weber, Mohamed Sallak. Data Uncertainty and Important Measures. Jean-François Aubry. ISTE Ltd and John Wiley & Sons Inc, pp.250, 2018, Systems and Industrial Engineering Series. Systems Dependability Assessment SET., 978-1-84821-993-9. ⟨hal-01683009⟩

Share

Metrics

Record views

145