The Individuals’ Learning Behaviors And Simulation In The Public Transit

The public transit riders’ learning behaviors are research subjects in this thesis. We have built a model to describle the riders’ choice behaviors when they faced the dynamics in the public transit system. Based on a Multi-Agents simulation system, we simulate the operational process of the public transit system and the learning and adaptive behaviors of riders. The emergent properties of the transit riders’ behaviors are also observerd in this system.We build a theoretical model framework of the public transit rider’s travel behaviors through the analysis of the individual’s cognitive factors and operational process of public transit. The properties of Markov Decision Process(MDPs) are expounded in this thesis. A public transit route choice model based on MDPs is formulated in this thesis. The algorithm of Reinforcing Learning is introduced to solve the above model. The perceived update and departure time choice models for public transit riders are formulated through the analysis of the rider’s learning mechanism. Through the simulation experiments, we found that the above models are appropriate for describing the individual’s learning behaviors and adaptation process.We analyse the role of information in the process of the individual’s cognitive learning, and build a mechanism of information evaluation and information reliable index updating. From the simulations,we find that information is an external factor in the travel. The travelers can become more sensitive by acquiring travle information. But they could not receive the extra benefits from information. The travle information can play a key role when there are great changes in the travel environment.We analyse the habit choice behaviours and the relation between the habit choice and cognitive learning in the public transit travel. A public transit travel choice model based on the individual’s habit choices is formulated. We also build a mechanism of habit choice triggered and terminated. From the simulations,we find that most travelers habit choice behavior is a useful factor for the system equilibrium in an uncertainty environment.Based on the SWARM platform, we have developed a multi-agent simulation system for conducting all the simulation experiments in this thesis.