Agent-based model for the rescheduling of Individual and collective daily activities under uncertainties
Modèle multi-agent pour la re-planification des activités individuelles et collectives quotidiennes dans un environnement perturbé
Résumé
Daily activity schedule are popular for people duringdaily life. While, when executing the schedule on the real road network, there are always some disruptions disturbing the planned schedule. To deal with this problem, daily activity rescheduling is necessary. This thesis regards the disruptions from the activity schedule execution environment as unexpected events (uncertainties). It establishes agent-based models to simulate the activity rescheduling decision process from the aspects of individual activity rescheduling and joint trip renegotiating.For the individual activity rescheduling, the model in this thesis wants to explore the relationship between a pair of episodes (two connected episodes) under unexpected events. Therefore, activity type is an important factor to consider. This thesis uses the decision tree to search all the alternative choices, and then it calculates the penalty after applying for the choices for each episode. For the joint trip renegotiating problem, when unexpected events happens, such as congestion, the driver and passenger need to renegotiate the drop off place and arrival time. The passenger drop off place may be a place near to his original location, or a new location. This thesis proposes a tolerance distance to find the alternative drop off place, and it uses the utility function to calculate the score of each alternative choice. Also, during the renegotiation process, this thesis considers the relationship between the passenger and driver, and also the time pressure. Both of them affect the person’s concession degree to his opponent.The goal of this thesis is to simulate the activity rescheduling decision and its focus is the travel behavior. It defines the unexpected events that may occur during the activity schedule execution process, and it establishes models to deal with both the individual activity rescheduling decision-making process and joint trip renegotiating process. It would like to provide a method to simulate the rescheduling decision-making mostly closed to the reality, while, it still needs to be validated to the real case in the near future.
Dans leur vie quotidienne, chaque personne realise un plan d'activités individuelles et collectives. Lors de l’execution de ce plan, certaines perturbations impliquent la nécessité de modifier le plan. Dans cette thèse ces perturbations sont assimilées à des événements captes par les individus. Une approche de modélisation multi-agent est utiliser afin de reproduire par simulation le processus de re-planification des activités tant du point de vue individuel que collectif. Dans ce cadre, la relation entre une paire d ́episodes (trajet et activité) successifs est étudié lorsque le plan est confronte à des perturbations. Cette thèse utilise un arbre de decision pour construire l’ensemble des planifications possibles, ainsi que leurs pénalités respectives pour chaque episode. Concernant la re-négociation des trajets partagés, le conducteur et le passager doivent re-négocier les lieux de deposes et l’heure d’arrivée, en minimisant la distance entre ces lieux. Ainsi, une distance de tolerance est utilisée pour trouver le lieu alternatif, ainsi qu’une fonction d’utilité pour calculer le score de chaque choix alternatif. De plus, au cours du processus de re-négociation, la relation entre le passager et le conducteur est étudiée, notamment du point de vue de la pression temporelle. Ceci affecte le degré de concession du conducteur et du passager vis-à-vis des propositions de son partenaire.
Origine : Version validée par le jury (STAR)
Loading...