Microscopic Modeling Of Vehicle And Pedestrian Traffc Based On Direction-changing Mechanism

Congestion of urban traffic has become a common problem in the urban development of China, and thus how to evacuate large crowd to safe area in limit time has become a serious problem, especially when disasters or accidents threaten public safety. The question of how to solve or alleviate the problem has been the most important in the work of traffic and public safety fields. To answer this question, computer modelling of the traffic and pedestrian flow to reveal their basic rule can be one of the important approaches, which can also provide scientific guiding principles for traffic or crowd evacuation under disaster environment.In this paper, the theme of our research is modelling the urban multi-lane traffic and two-dimensional pedestrian movement; the aim is to build microscopic models that can objectively reflect the basic characteristics of vehicle and pedestrian flow, and the basic idea is to build a two-dimensional continuous lane-changing vehicle model and pedestrian model with the unified model framework, which develops the one-dimensional car-following model. Thus we achieve the extension of the vehicle traffic model and improvement of the the pedestrian dynamics model.To explore the general mechanism and modeling method of the direction change of traffic individuals, in Chapter2, we firstly analyzed the lane-changing rules of the typical multi-lane models and the direction-changing mechanism of the typical pedestrian movement models by literature investigation. Subsequently, similar characteristics between vehicle traffic and pedestrians were summarized as the behavioral process of changing movement states by human. We concluded such process as four stages, namely demand generation, environment assessment, decision making, and program implementation, and then proposed that modelling the lane changing or direction changing is essentially the process of modelling the four stages.In order to develop lane-changing models that can objectively and truly reveal the characteristics or rules of urban traffic flow, in Chapter3, we proposed three lane-changing models, respectively, according to the speed limit of lane, the concurrent of discretionary and compulsive lane changing before traffic bottleneck, and the time-process character of lane-changing process. We also carried observation experiment of multi-lane traffic and validated our model by comparing the simulation and empirical data. Results show that:1) the essential reason for the "density inversion" phenomenon is some factors exist to induce asymmetric lane-changing behaviour;2) the lane with high speed limit would lose its velocity advantage and becomes "faster but not fast";3) the frequent lane changing in the medium or high traffic density would reduce the total traffic flow of the roadway and make it "haste makes waste";4) the traffic bottleneck would promote the lane-changing behaviours of the vehicles upstream, reduce the velocities of the vehicles in a large range of the upstream roadway and delay the travel time of traffic individuals;5) as the traffic density increases, the lane-changing frequency firstly increases and then decreases, and the maximal frequency occurs in the density range15～35vehs· km-1· lane-1.In order to develop two-dimensional pedestrian model that can realistically and accurately reflect the characteristics or rules of pedestrian flow, in Chapter4, we referred the modeling idea and method of multi-lane traffic to modeling pedestrian movement. We analyzed the single-file experiment to obtained the expected velocity function of pedestrian, and then solve the issue of velocity adjustment in the pedestrian’s movement by means of combining the FVD acceleration equation. We carried obstacle avoidance experiment to extract the direction-changing probability function, and then solved the issues of direction-making and direction-choosing through integrating the proposed "visual hindrance". Based on the mechanisms of velocity adjustment and direction changing, we presented a new continuous pedestrian model that rely pedestrian’s visual information. And we also validate the proposed model by comparing our simulation and the empirical data derived from experimental studies conducted by domestic and overseas researchers. Results show that:1) there is a strong correlation between the velocities of one pedestrian and her/his front adjacent pedestrian, and the pedestrian has a tendency to keep synchronous velocity with her/his front adjacent pedestrian;2) the direction changing of pedestrian is probabilistic, and the probability increases with the decreasing distance to the obstacle;3) the pedestrian model considering "visual hindrance" can reasonably and accurately describe the characteristics and rules of pedestrian flow;4) the vehicle traffic model framework of FVD is applicable to both the multi-lane modeling and the two-dimensional pedestrian movement modeling.At last, in Chapter5, we summarized this modeling study and presented the principal conclusions. Meanwhile, further research in this work was stated.