Modeling and Simulation of Travel Behavior Forecasting System Based on Bayesian Network

We establish a forecasting system for travel behavior, which is aimed to analyze the attributes of residents travel decision-making behavior. We study the relations between various factors which affect activities and travel decisions by different strength and order. The forecasting system, with a framework based on Bayesian theory and activity-based travel behavior analysis method, include the input layer module, the simulation layer and the output module. Establishing the Bayesian belief network with the activity-based method, we focus on analyzing the connection of activity generation module and activity arrangement module under the simulation layer. We make predictions about individual travel behavior, which is the response to transport policy change, and analyze the travel decision-making chain structure through the output layer. With the understanding of the logical relations among various modules, we work out the simulation processes. The results show that the Bayesian forecasting system can analyze multi-trip travel decision-making behavior and the linkage among them.