Transférabilité d'une modélisation-simulation multi-agents : le comportement inter-gares des voyageurs de la SNCF lors des échanges quai-train

Abstract : This thesis focuses on the definition and the validation of a crowd-simulation modeling method. This method allows transferring principles and measurements established in a source observation context to a target one. From the observation of pedestrian behavior in a social and urban context, our approach infers rules and motivations in order to model and simulate the individual pedestrian behavior in this source context. This allows, in a second step, to evaluate the transferability of the model to a given target context. Our work is based on a real case: the simulation of the passenger’s behavior while boarding and alighting trains (platform-train exchange) in a high-traffic station (dense situation). Starting from the state of the art of theories, models and types of simulation, we choose an approach based on the microscopic simulation of social agent’s motivation. This choice relies mainly on the fact that, compared to a classic crowd simulation, it has, a priori, a better transferability potential. The goal of this type of simulation is to study the architectural evolutions of the stations that could allow to control or shorten train dwell times.
Document type :
Theses
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-02137524
Contributor : Abes Star <>
Submitted on : Monday, June 24, 2019 - 9:35:07 AM
Last modification on : Tuesday, June 25, 2019 - 1:47:45 AM

File

TheseELLEUCHdiff.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-02137524, version 1

Citation

Fatma Elleuch. Transférabilité d'une modélisation-simulation multi-agents : le comportement inter-gares des voyageurs de la SNCF lors des échanges quai-train. Intelligence artificielle [cs.AI]. Conservatoire national des arts et metiers - CNAM, 2019. Français. ⟨NNT : 2019CNAM1227⟩. ⟨tel-02137524⟩

Share

Metrics

Record views

197

Files downloads

67