Skip to Main content Skip to Navigation
Theses

Traitement ordinal de l'information d'expertise pour le risque en génie civil : apport des sciences de la décision à la gestion des risques

Abstract : When facing high uncertainty systems, such as dams, where experts heuristics becomes too much important, usual tools are not satisfying enough to reveal experts’ opinion in order to manage the risks associated with the system. Decision science then brings tools to sharpen our understanding, or even help the elicitation, of what the expert wants best to express. Concerning dams, we have only very little feedback, and no to few significant events. In addition to the lack of knowledge when it comes to the phenomenological mechanisms, these issues lead us to use unusual tools for risk management. This study brings an innovative tool to help on the elicitation of experts’ opinion, allowing risk management on dams based on an ordering approach. Furthermore, we will show this tool is an estimation of the maximum likelihood, which is invaluable information for any decision maker. We will show this result is obtainable through a method using rule based assignments, developing the rules thanks to tools like votes, games and database theories. Doing so, we will show how it is possible to process risks without using usual probabilistic tools, while taking experts’ heuristics into account.
Document type :
Theses
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-02089172
Contributor : Abes Star :  Contact
Submitted on : Wednesday, April 3, 2019 - 3:07:41 PM
Last modification on : Sunday, October 25, 2020 - 7:04:32 AM

File

TORET.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-02089172, version 1

Citation

Jean-Baptiste Toret. Traitement ordinal de l'information d'expertise pour le risque en génie civil : apport des sciences de la décision à la gestion des risques. Gestion et management. Université Panthéon-Sorbonne - Paris I, 2014. Français. ⟨NNT : 2014PA010012⟩. ⟨tel-02089172⟩

Share

Metrics

Record views

138

Files downloads

99