| 英文摘要 | The last few years have witnessed the problems of urban transportation becoming more and more serious. Most traffic departments have been sparing no efforts to seek the solutions that can alleviate congestion and reduce accidents. Currently, one of the effective methods is appropriately increasing the traffic supply and taking rational measures to restrict or guide the travel demand dynamically. As a fundamental part of travel demand forecasting, the disaggregate analysis of urban residents' travel behavior has attracted more and more academic attention. Studies on travel behavior can not only provide the theoretical basis of urban transportation planning in order to increase traffic supply, but also develop credible models for simulation and computational experiments so as to evaluate a variety of traffic management schedules. Most of the existing theories about disaggregate behavior focus on the economy, environment, social status, and psychological characteristics of the individuals, and do not take the actual traffic state into account. So there are a lot of restrictions when applying these theories. Based on the ACP (Artificial systems, Computational experiments, Parallel execution) approach, a theory about the modeling, analysis, management, and control of complex systems, this thesis studies the parallel calibration method for travel behavior of artificial populations, so that the traffic phenomena ``emerged'' from particular travel mechanism can approximate the actual system. Specifically, the calibration involves three elements: destination, departure time, and route. Moreover, the travel calibration module of artificial populations is designed and implemented and the experiments are carried out in the existing TransWorld software---a platform for computational experiments in Artificial Transportation Systems. The main content of this thesis is as follows: 1. A comprehensive literature review on the theory of disaggregate travel behavior is presented, including random utility, destination selection, departure time selection, and route selection. The advantages and some problems of the existing research results are also analyzed. 2. A computation method of OD (Origin-Destination) vector based on compressed sensing is proposed. Limited by timeliness and measurement scale, dynamic OD estimation in urban transportation can only be calculated indirectly in the case of under-sampling, which brings much deviation to the results. In this thesis, a new... |
修改评论