基于流体力学模型和Unity引擎的城市交通流仿真平台

The mechanism of traffic flow has been a controversial issue for decades. This in our view is mainly because (i) the modeling and simulation focus on a single road more than a road network, and (ii) the data or observations are inadequate and less relevant to a road network. On the other hand, applications of dynamic transportation assignment (DTA) are calling for precise description of traffic flow on road networks, and computer and information technologies bring about the revolution for studies, managements and controls of traffic flow concurrently with large amounts of traffic data. These stimulate the development of the traffic flow theory on road networks and software both for the study and application. .In the context, this research project develops a simulating platform for urban road networks by using Unity Engine. The platform includes four program modules for (i) reading and editing the network information; (ii) designating the model and network parameters; (iii) simulating traffic flow on the network; and (iv) analyzing the outputs (tables, figures and visualized images). The software SketchUp for architecture design is also applied to generate scene graphs in high quality. Moreover, it seems to be the first time that we apply the hydrodynamics traffic flow model for the simulation on such a platform. Since the hydrodynamics model is macroscopic, which suggests much efficiency for computation, data storage, and visualized display, the developed platform would be suited for most cities in China, which are characterized by huge size or/and complex urban road networks. .In the first phase of the project, the four program modules together with the platform will be developed based on an imaginary or abstract network, which is mainly for the study of traffic flow mechanism, especially through combination with DTA problems. In this regard, the platform could be viewed as an updated version of the current simulation tools. .In the second phase, the platform will be extended to mimic traffic flow in some Chinese cities (e.g., Beijing and Shanghai). For the purpose, we will abstract the main features of the city traffic from taxi or other mobile data. This includes the following: (i) generating the road network from traffic trajectories of the taxis; (ii) obtaining the messages regarding the origin-destination (O-D) pairs, path-choice and traveling behaviors, etc; and (iii) adapting the models and parameters to the data..In the third phase, efforts will be taken to connect the platform to those traffic monitoring centers favorably in the studied cities. Thus, the platform is expected to serve for urban transportation planning and control, through the following: (i) publishing traffic indexes (jamming, delay, pollution, etc); (ii) evaluating transportation planning (e.g., signal control and network design); (iii) making online and mid-long term forecasts of traffic flow.

针对大型和复杂路网，运用Unity引擎自主研发适于我国城市交通流演化的流体力学模型仿真平台。平台的实现将主要基于对以下四个功能模块的开发：（i）用于读取和编辑交通路网的路网建模模块；（ii）用于设置路网和模型参数的参数设置模块；（iii）用于模型数值计算的模型仿真模块；（iv）用于可视化和仿真结果分析的结果分析模块。平台的完成将是对已有交通流数值试验手段的升级，并能结合交通分配问题中的O-D关系和路径选择原理，直观、动态和全面地展现城市交通流的数据、图表和图像。研发内容将针对我国的具体城市，运用其出租车和通讯等移动数据，通过模型参数识别、生成城市交通路网和提取O-D信息等手段，对城市的交通设计或改进以及信号灯控制等方案做预判；将力图实现与交通监测中心的对接，为城市交通管理和控制提供实时和在线服务，发布各种交通指数，尤其是做短、中和长期的交通流宏观预测。

我国交通具有不同机动车混合、机动车-自行车-行人之间相互干预、交通行为不够规范以及城市网络相对复杂等特征。因此，自主研发交通仿真平台是解决我国城市交通问题的必然选择。本项目自主开发出较为完整的Windows桌面版和网页版交通仿真平台，实现了场景搭建、参数设置、模型仿真和结果可视化等功能，涵盖道路交通流、网络交通流和行人流等多个方面。研究内容还包括与交通仿真平台密切相关的交通车流与行人流理论建模和数值方法。.针对项目提出的有关城市典型交叉口的交通流建模和仿真需求，通过对城市交叉口（丁字路口、环形交叉口）转向区域进行了元胞交错划分，合理假设了模型的随机慢化函数，提出了改进的元胞自动机交通流模型，并实现了模型的动态仿真模拟。.针对项目提出的有关实际交通路网中道路不均匀情况建模和仿真需求，建立了流通量间断的守恒高阶（CHO）交通流模型并实现了数值模拟。深入研究了高阶交通流模型的理论性质，并运用泰勒展开法和约化摄动法给出了模型的近似孤立波解析解。.针对项目提出的有关交通路网建模和仿真的需求，将LWR模型推广到了交通路网，较为完整地讨论了在用户均衡条件下了LWR模型的单OD路网定常解，为建立包含激波间断的静态交通分配模型提供了关键思路和理论依据。.针对项目提出的有关城市宏观交通流建模和仿真的需求，提出了关于城市交通流的二维连续型流体力学新模型，包括（1）同时考虑出行时间和路径选择的动态交通分配模型；（2）考虑住房分布的二维动态交通分配模型；和（3）城市交通与土地利用耦合的二维连续型双层规划模型。.针对项目提出的有关复杂场景下的行人流建模和仿真需求，建立了基于势场的细化网格元胞自动机模型，对雨天、高密度等复杂场景下行人疏散、对流和挤压等现象进行了数值模拟和仿真。.综上所述，本项目基于先进的交通车流和行人流建模和仿真理论，通过自主研发建立了较为完整、安全和可靠的交通仿真平台。该平台既可作为交通科学数值实验的可视化工具，也可与大数据等信息技术相结合，进一步应用于交通监测与仿真模拟。基于所研发的交通仿真平台，我们还与交通管理部门和交通领域高科技企业合作，开展了城市路网提取、多源数据（GPS、视频和百度地图）分析、交通状态估计与预测等工作，为将所开发的仿真平台应用于城市交通或城际路网做了有益的准备。因此，项目开发的交通仿真平台不仅具有重要的科学意义，也有实际的应用前景。