Urban freeways play an important role in urban traffic networks. Compared with highway traffic, urban freeway traffic has different characteristics, such as denser on-and off-ramps, more complex road conditions, and lower velocity limits. Until now, however, there has been no comprehensive analysis of urban traffic flow. In this paper, through an analysis of the density dependence of velocity distribution, we investigate the fundamental velocity–density relationships of urban freeways, compare them with those of highway traffic, and explain them using existing traffic flow theories.

At present, research on traffic flow theory has mainly focused on highway traffic, which is significantly different from urban freeway traffic. Traffic science is an "empirical science", and as such is it is based on empirical urban freeway traffic flow data. In this study, first four steady phases are identified in the flow-density plane of traffic flow: free flow, coherent-moving flow, synchronized flow and jam. Then, three modes of phase transition, i.e. spontaneous transition, propagating transition (PT) and induced transition (IT) are analyzed, and the judgment conditions as well as computable methods of PT and IT are discussed in detail. Finally, based on actual data of the Beijing urban freeway, an empirical analysis of phase transition from coherent-moving flow to synchronized flow, which is the process of "smooth traffic" transferring to "congested traffic", is presented to validate the concepts and methods proposed in this paper.

公共交通线路调整影响分析是一个定量与定性相结合、经验与理论相结合、创新与反馈相结合的复杂过程，是一个要涉及协调多方面利益和目标的寻优过程。基于四阶段交通模型理论，使用O-D反推技术和随机用户均衡配流模型，设计了公共交通线路调整影响分析流程，达到更实际、更合理的交通分配结果。以8图层架构为基础，提出了系统总体架构，最后使用TransCAD实现系统。通过某市实际的公交路网的模拟计算，得到了合理的结果，证明了系统具有较好的应用价值。

Bottlenecks at highway or urban freeway are one of origins of traffic congestion. Based on empirical investigation on field data of Beijing urban freeway, schematic phase diagram at on-ramp vicinity is developed and detailed phase transition processes are investigated. There are five phases in the phase diagram of on-ramp vicinity in connection with on-ramp flow and upstream flow, among which are two phases only observed in urban freeway traffic. It's also found that the empirical phase diagram developed in this paper is different from previous theoretical phase diagram of highway traffic. The boundaries of each phase are influenced by traffic demand at locations, such that empirical phase diagrams are vary in quantities at different locations.Moreover, according to investigation on empirical phase transition process, coherent moving flow, as a new phase observed in urban freeway flow in our previous work, is confirmed.

目前交通流理论研究的主要对象为高速公路，与其相比，城市快速路交通流则具有明显不同的特点。交通科学是一门说证科学，在实证数据的基础上，首先对城市快速路交通流速度概率分布进行了研究，分析了在不同密度条件下速度概率分布的特性。在此基础上，在流—密平面中将交通流划分为自由流、谐动流、同步流和堵塞4个稳态相位，并具体讨论了各个相位的性质，定量标定了每个相位的区域范围。

在介绍现有的主要交通流预测方法的基础上，阐述了基于卡尔曼滤波（Kalman）的预测模型及其具体算法。结合城市环路的交通运行特性，构建了基于卡尔曼滤波的交通流短时预测模型，并根据北京市三环路的实际数据对模型进行验证。实证数据表明，所建立的交通流动态实时预测模型的预测效果比较理想，算法的实时性也满足实际预测系统的要求，可应用于交通流预测及交通智能控制。

交通系统的状态转移是一个正负反馈交替发生作用的过程。首先提出了同态度量指标、时空差异度指标等新概念用于定性判断交通系统的状态，并在此基础上，建立了一组变参数线性方程用以描述交通系统状态转移的正负反馈交互迭代的过程。最后以某城市环路一个路段发生交通拥堵的过程为实例，仿真分析了交通拥堵发生时车道非同态性变化的过程，仿真结果表明提出的模型能较好地反映车道非同态性变化的实际情况。

基于BP神经网络和RBF神经网络构建了城市环路交通流动态实时预测模型，并对北京市三环路的实测交通流数据进行了预测分析。实证结果表明，该模型具有较高的精确度，算法的动态实时性也满足实际预测系统的要求，具有较好的实用性。

At present, researches on traffic flow theory mainly focus on highway traffic, which is significantly different from urban freeway. Because traffic science is an "empirical science", it is based on actual traffic flow data of urban freeway, the statistical features of velocity distribution at different densities have been investigated in this paper. Moreover, four steady phases: free flow, coherent-moving flow, synchronized flow, and jams are identified in the flow rate-density plane of the traffic flow. Finally, features of the four phases as well as functional forms of phase boundaries are discussed in detail.