城市快速路瓶颈路段交通拥堵时空演化及实时估计方法

Urban freeway possesses many characters in China, ramp distances are near, correlation is very strong between freeway and road network, lane changing occurs frequently. Purpose of this research is to explore the mechanism of traffic congestion formation and evolution in the Recurring Bottlenecks including merge area, diverge area and weaving section and so on, to propose a scientific method to estimating the real-time traffic state at these bottlenecks. Firstly, this research will collect the spatio-temporal vehicle trajectories at the bottlenecks, propose the probability of lane-changing based on traffic density distribution across the lanes, bottleneck location and driver character. Obligatory lane change factor is introduced to express the effect of obligatory lane-change maneuver, considering the fact that obligatory lane-changing occupies more spatio-temporal source. Secondly, microscopic lane-change behavior is planted into the traffic dynamical model. Starting with the multi-lane macroscopic traffic model, traffic model for bottleneck is built, considering several factors including Obligatory lane change factor, ramp volume distribution, bottleneck combination mode and correlation between freeway and road network. Simulating experiment will explore the traffic congestion formation and evolution under various influencing factors.Thirdly, spatio-temporal correlation was investigated in the traffic flow. Measurement error will be analyzed for different traffic flow detecting techniques. One real-time traffic state estimation method will be proposed combining the macroscopic traffic model and Kalman filter. Fruit of this research will help us explore the mechanism of congestion evolution , alleviate the degree of congestion on urban freeway.

城市快速路具有匝道间距短、与地面路网关联性强、换道操作频繁等特征。本研究围绕快速路合流区、分流区和交织区等常发性瓶颈路段交通阻塞形成演化机理、交通状态实时估计展开研究。采集车流轨迹时空数据，结合瓶颈段车流密度分布、瓶颈位置、驾驶员特性建立换道概率方程；针对强迫式换道占用额外时空资源的物理现象，引入强迫式换道影响因子表征其对交通流作用。将微观换道行为植入交通流动力学模型中，以多车道宏观交通流模型为基础，结合强迫式换道影响因子、匝道流量分布、关联瓶颈组合模式、与地面路网关联等因素，构建瓶颈段的交通流模型，通过仿真实验模拟各因素影响下交通流的时空演化特征。开展交通流时空关联性研究，从采样周期、数据颗粒等研究各类交通流检测技术的测量误差，结合交通流模型和卡尔曼滤波技术提出快速路交通状态实时估计方法。本研究对于科学解析快速路交通拥堵演化机理、减轻快速路拥堵程度具有重要的理论和实用价值。

城市快速路是城市道路交通系统的主动脉，由于快速路的特性，如：匝道间距短、匝道间关联性强、主线和匝道流量均很大、且强制合流、强行换道等，在快速路上拥堵时常发生，且拥堵发生后会快速向上下游扩散，这给交通管控带来巨大挑战，快速路常发性瓶颈拥堵演化是国内外长期关注的课题。本课题主要从三个方面对城市快速路瓶颈路段交通拥堵时空演化及实时估计方法进行研究，首先，以非对称驾驶行为理论为基础，运用停走交通波揭示拥堵产生及扩散的机理，将停走交通波的演变进一步细分为增长波或消散波，并提出了判别方法，定量分析了变道对停走交通波以及交通状态的影响；其次，基于改进的元胞传输模型的城市快速路交通流时空演化仿真模拟，并针对入口匝道流量对主线车流的影响和出口匝道交叉口通行能力对主线车流的影响进行了分析，提出了缓解交通拥堵的匝道控制方法，通过仿真模拟验证了该方法的有效性；最后，提出了基于多源数据融合的快速路交通流状态估计方法，选用极限学习机模型，对多源数据进行融合处理，以获得更精确的交通信息；对其进行时空匹配，从而获得同一时间和同一空间的交通状态多源数据；解决了采用单一的数据难全面准确地估计路网实时状态的问题。本研究对于科学分析快速路交通拥堵演化机理、通过交通流状态估计和匝道控制等方法，缓解城市快速路拥堵具有重要的理论意义和实用价值。