Shockwave-based queue estimation approach for undersaturated and oversaturated signalized intersections using multi-source detection data

With the progress of information and sensing technologies, estimating vehicular queue length at signalized intersections becomes feasible and has attracted considerable attention. The existing studies provided a solid theoretical foundation for the estimation; however, the studies have some restrictions or limitations more or less. This paper presents a new methodology for estimating vehicular queue length at signalized intersections using multi-source detection data under both undersaturated and oversaturated conditions. The methodology applies the shockwave theory to model queue dynamics. Using data from probe vehicles and point detectors, analytical formulations for calculating the maximum and minimum (residual) queue lengths of each cycle are developed. Ground truth data were collected from numerical experiments conducted at two intersections in Shanghai, China, to verify the proposed methodology. It is found that the methodology has mean absolute percentage errors of 17.09% and 12.28%, respectively, for maximum queue length estimation in two tests, which are reasonably effective. However, the methodology is unsatisfactory in estimating the residual queue length. Other limitations of the proposed models and algorithms are also discussed in the paper.     

Probe vehicle lane identification for queue length estimation at intersections

Vehicles instrumented with Global Positioning Systems, also known as GPS probe vehicles, have become increasingly popular for collecting traffic flow data. Previous studies have explored the probe vehicle data for estimating speeds and travel time; however, there is very limited research on predicting queue dynamics from such data. In this research, a methodology was developed for identifying the lane position of the GPS-instrumented vehicles when they are standing in the queue at signalized intersections with multiple lanes, particularly in the case of unequal queue. Various supervised and unsupervised clustering methods were tested on data generated from a microsimulation model. Among the tested methods, the Optimal Bayes Rule that utilizes probability density functions estimated using bivariate statistical mixture models was found to be effective in identifying the lanes. The methodology for lane identification was tested for queue length estimation. This research confirms that the lane identification is an important step required prior to the queue length estimation. The accuracies of the models for lane identification and queue length estimation were evaluated at varying levels of demand and probe vehicle market penetrations. In general, as the market penetration increases, the accuracy improves as expected. The result shows that 40% market penetration rate is adequate to reach about 90% accuracy.     

基于抽样车辆轨迹数据的信号控制交叉口排队长度分布估计

排队长度是评价信号控制交叉口运行状态的重要参数之一。现有大多数基于抽样车辆轨迹数据的排队长度估计方法可以实现周期级排队长度估计，但是需要信号配时、渗透率或车辆到达分布等实践中难以获取的输入信息。此外，这类方法在低渗透率条件下往往难以确保估计结果的准确性和可靠性，极大地限制了其实用性。因此，提出一种抽样车辆轨迹数据驱动的时段级信号控制交叉口排队长度分布估计方法，可不依赖任何交通流理论模型和前述输入信息实现排队估计。首先，通过理论推导可以证明时段内抽样车辆的停车位置分布和排队长度分布之间可互相转化；然后，提出一种扩展的核密度估计方法来拟合并平滑抽样车辆停车位置分布，从而有效地适应不同日期和周期的轨迹叠加所带来的波动，提高方法的适用性；最后，基于前述推导和拟合的停车位置分布实现时段排队长度分布、平均排队长度和百分位排队长度估计。分别采用仿真和实证数据对上述方法进行验证和评价。结果表明，通过叠加5 d相同时段的抽样轨迹数据，15 min的平均排队长度估计误差仅为1.59 veh，相对误差仅为9%。同时，面向不同分析时长，只要给定超过100 veh抽样车辆的观测样本，无论渗透率高低，所提出的方法在定时或自适应信号控制交叉口都可实现时段排队长度分布的准确估计，其成果可进一步用于信号控制交叉口运行可靠性评估以及多时段定时信号控制的鲁棒优化。     

基于网联车辆轨迹数据的周期排队长度估计

近年，基于网联车辆轨迹数据的交通管控与服务研究方兴未艾。其中，信号控制交叉口排队长度估计备受关注。然而，在低渗透率条件下，单个周期内轨迹稀少且提供的交通信息十分有限。现有研究仅以当前周期内网联车辆轨迹数据为输入，难以获得准确且可靠的周期级排队长度估计结果。因此，融合利用历史网联车辆轨迹数据提供的车辆到达和停车位置信息以及当前周期内实时观测的网联车辆排队信息，提出一种基于最大后验概率的周期最大排队长度估计方法。首先，依据历史轨迹数据的停车位置信息，估计排队长度的先验分布；其次，依据历史轨迹数据的车辆到达信息，估计周期内车辆的历史到达分布，并结合周期内最后1辆排队网联车辆的到达时刻与停车位置，构建排队长度似然函数；最后，基于贝叶斯理论，结合前述先验分布与似然函数，推导周期排队长度的后验分布，并采用最大后验概率方法实现周期最大排队长度的估计。仿真结果表明：所提方法在不同饱和度和渗透率条件下，均优于现有的方法；即使在车辆轨迹数不超过1 veh·周期^-1的低渗透率条件下，所提方法的平均绝对估计误差也不超过2 veh·周期^-1。实证结果表明：在渗透率仅为8.96%的条件下，所提方法的平均绝对误差为2.12 veh·周期^-1，平均相对估计误差为12.4%，同样优于现有同类方法。     

临界饱和状态交通干线协调控制模型

为提升临界饱和状态下干线车流通行效率，提出了一种基于冲击波理论的干线双向信号协调控制方法。首先，建立了考虑车速变化、转向比例、车道变化等因素的干线交通流模型，分析了临界饱和交通干线交通流运行状态与各参数间的关系。第二，构建了以干线双向通过量最大化为优化目标的混合整数线性规划模型，通过优化干线公共周期和各交叉口绿信比以提高干线通行能力。第三，构建了以延误最小化为优化目标的二次规划模型，并提出了相应的求解算法，通过优化相位差和相位方案实现了干线交叉口的信号协调。临界饱和交通干线协调控制模型由通过量最大化模型和延误最小化模型构成，考虑各交叉口间的制约影响关系，有效避免了排队滞留、溢出、交叉口“死锁”等现象。采用两阶段优化方法，通过通过量最大化模型优化周期及绿信比，继而应用延误最小化模型优化交叉口相位方案及相位差，获得干线系统双向信号协调最优控制方案。最后，应用临界饱和交通状态干线协调控制模型对南京市中山东路10个交叉口进行了信号协调优化，并对优化结果进行了仿真分析。结果表明：临界饱和交通状态干线协调控制模型能对双向临界饱和干线的信号控制方案进行优化，与对照方案相比，优化方案的双向总交通量提升了21.9%，车均延误降低了63.1%，通行能力与服务水平提升显著。     

基于单截面低频检测数据的信号交叉口排队长度估计

针对我国大多数中小城市信号交叉口交通检测数据条件及基于此数据条件下存在的信号交叉口排队长度估计精度不高问题,研究了基于单截面低频定点检测数据的信号交叉口排队长度估计模型.利用时间占有率与流量、速度之间的函数关系对长排队(排队长度超出检测器位置)进行识别.根据信号配时数据切分低频检测器数据,并与信号配时数据匹配.基于交通波理论,通过关键点的判别求取周期最大排队长度.以青岛市山东路-江西路南进口为例进行仿真和实证验证.结果显示,长排队的识别精度达到了90% 以上,不同饱和度下(低、中、高)的信号交叉口排队长度估计精度均达到了80% 以上,其中,中、低饱和度场景下排队长度平均绝对误差小于20 m/cycle,高饱和度场景下排队长度平均绝对误差小于45 m/cycle.      

基于交通流模型的道路权重计算方法研究

提出了一种基于交通流模型的计算道路权重方法。实施定点道路截面雷达试验并采集交通流信息,标定生成交通流速度-密度模型。通过此模型,利用出租车行驶数据库中各条道路车速数据计算各道路车流密度,根据各道路长度和密度推算道路权重。     

Platoon recognition using connected vehicle technology

This article presents a mathematical model for real-time platoon recognition using the connected vehicle (CV) technology. Platoon information is a crucial part of traffic signal coordination and is difficult to obtain with traditional technologies such as loop detectors. The past work on platoon recognition using CV is very limited and lacked verification on the applicable range or evaluation of the performance of algorithms. The proposed algorithm is focused on estimating platoon characteristics for signal coordination and adaptive signal control with CV's vehicle-to-vehicle communication and an onboard GPS device. First, the detected platoon is identified by a modified critical time-headway. Then, platoon size and starting and ending times are estimated. Lastly, the filtering process for “qualified” detected platoon is proposed to optimize detectability. The results show that the proposed algorithm can estimate well in various traffic conditions and under both fixed-time and actuated signal control without the need for recalibration. Furthermore, two analytical models to estimate the detection rate are proposed and shown to be close to the numerical results and can be used to estimate the required market penetration ratio for the application without field experiments or microscopic simulation. The accuracy of both the recognition algorithm and detection rate estimation is obtained without relying on inputs that are hard to obtain in practice. Accordingly, the proposed algorithm can be an important part of adaptive signal control focusing on real-time coordination in CV environment.     

综合客运枢纽行人集散设施瓶颈识别方法

为充分发挥综合客运枢纽行人集散设施的能力,提高枢纽运行效率,分析枢纽行人集散设施瓶颈的形成机理,给出瓶颈的定义和识别流程,提出客流拥堵自动识别算法,考虑拥堵强度、拥堵持续时间和拥堵频率等拥堵特性,构建综合客运枢纽行人集散设施瓶颈识别指数模型并制定瓶颈分级标准.最后以北京南站地下一层换乘大厅为例,验证了方法的有效性.结果表明,高峰时段换乘大厅地铁进出闸机处为Ⅰ级瓶颈,平均排队长度5人,地铁东北、西南售票处为Ⅲ级瓶颈,平均排队长度为14人,与实际情况相符,而通过瓶颈缓解措施,能有效降低排队长度.      

高架出口匝道下游交叉口信号控制方法研究

城市高架快速路与地面道路,主要通过出口匝道及其下游交叉口进行交通转换,高峰时段出口匝道及下游交叉口交通拥堵频发。以元胞传输模型为基础,构建出口匝道及下游交叉口交通预测模型;采用动态调整周期时长和信号相位的控制策略,建立基于元胞传输模型的交叉口信号控制模型。以排队长度、绿灯和周期时长为约束条件,以各进口道加权平均延误为目标函数,进行信号配时动态优化。以成都市实例匝道和交叉口进行验证,表明本文提出信号控制策略可有效降低此类交叉口的饱和度、延误和排队长度,提升其通行效率。     

基于概率模型的交叉口感应信号控制研究

考虑多种因素影响下初始绿灯时间和车辆到达服从概率分布特征,提出了1种交叉口感应控制最大绿灯时间优化模型,以交叉口平均延误、停车次数、排队长度为评价指标,利用 Vissim 仿真软件,将该模型与传统感应控制模型进行对比。仿真结果表明提出的算法在不同的饱和度下改进效果比较明显,尤其当交通量较大、饱和度较高时,传统感应控制效果有限,采用该模型的改进效果依然显著。      

The ‘safety in density’ effect for cyclists and motor vehicles in Scandinavia: An observational study

Safety in density (SID) potentially explains the safety in numbers (SIN) phenomenon by positing that ‘the SIN effect can be reproduced simply through encouraging behaviour that leads to the formation of higher-density cyclist groups’. The study further explores this hypothesis using event-based exposure, queues and groups of road users. Using three different definitions of encounters between road users, these were manually counted at signalized intersections, and their relationship to traffic volume was assessed. Based only on the frontmost motor vehicle in a queue and one cyclist among the several passing in front of that vehicle, the results show a less than linear relationship between meetings and traffic volume. An increase in the number of cyclists entails a general increase in cyclists passing in front of each motor vehicle, and an increase in motor vehicles increases queue lengths. However, crash data from the Swedish accident database (STRADA) show that it is exceedingly rare for multiple cyclists to be injured in the same crash. Together with results from a crash-encounter model, this suggests that the SID hypothesis may help to explain SIN     

Identifying Time-of-Day Breakpoints Based on Nonintrusive Data Collection Platforms

To ensure the effective operation of traffic signal systems, different signal timings should be designed to accommodate traffic pattern variations. One of the greatest challenges is the identification of appropriate time-of-day (TOD) breakpoints, where different signal timings could be implemented during the time periods between two consecutive breakpoints. This research presents an advanced cluster analysis aimed at identifying TOD breakpoints for coordinated, semiactuated modes when it is necessary for multiple intersection operations to be considered simultaneously. Different from previous studies, this proposed methodology considers the time of traffic occurring as one dimension in clustering and uses continuous traffic data obtained through innovative, nonintrusive data collection techniques, which significantly improve this method's performance. The operability of this proposed method is demonstrated in a case study of a corridor located in Tampa, FL. The traffic simulation results reported in this article reveal that this novel procedure performs better than existing TOD signal timing plans.     

单点交叉口最优信号周期模型研究

为了实时选择最优信号周期,以城市道路单点交叉口为对象,分析了最短信号周期与最佳信号周期的推导过程,由此确定了4种信号周期的计算模型。采用延误、最大排队长度、平均停车次数和通行能力4个指标对4种信号周期模型进行综合评价,并确定出最优信号周期,以弥补评价指标少的局限性。在评价过程中,对不同指标的量纲范围进行了标准化处理,同时确定了不同指标的权重系数。通过实际数据和mat-lab编程对综合评价模型进行验证。结果证明,采用单一指标选择信号周期,会有不同的结果,且会造成其他指标不理想;采用综合评价模型可以直观地确定出最优信号周期,并保证交叉口的4个指标达到整体最优。      

基于出站时间的高速公路收费站车辆排队检测

目前,对高速公路收费站进行服务评价的数据主要依靠人工现场调查或使用专用检测设备来获取,在人力、资金方面消耗较大。而收费站的海量收费数据则可通过自动调查直接获得,充分利用这些收费数据资源可以获得很好的经济价值。分析了目前国内在公路收费站排队研究方面的局限性和不足。基于高速公路收费站的实际数据,研究分析了排队情况下同车道相邻车辆的出站时间规律,对收费站的车辆排队检测算法,以及车辆队长、排队逗留时间、服务时间等指标的量化计算进行了方法设计。以陕西省富平、韩城和芝川收费站的收费数据进行了实例计算,并与 M/G/1模型的计算结果以及调研数据进行对比,分析其绝对误差,对比结果表明平均服务时间的绝对误差不超过3 s;除韩城收费站105车道外,平均队长的绝对误差不超过1辆;除富平收费站101车道外,平均逗留时间的绝对误差不超过10 s。验证了算法的有效性与普适性。      

船舶航行交通事件实时检测技术研究现状与展望

船舶航行交通事件检测依赖基于历史数据的离线检测方法, 检测模型适用性差, 难以满足监管人员的实时监测需求。通过分析船舶异常行为检测、航行事故检测等现有交通事件检测技术, 可以发现: 在数据层面, 监测数据来源单一、环境信息缺失; 在方法层面, 基于统计、风险评估等经典模型的事件监测方法效率高但准确性低, 基于神经网络、图像识别等机器学习的检测方法准确性高但效率低; 多源数据融合、多项技术结合的交通事件检测方法成为实时检测方法的发展趋势。在此基础上, 梳理了实时船舶航行交通事件检测的3项关键技术: (1)海事大数据技术: 高效处理船舶运动数据和航行环境数据, 统一多源异构数据结构标准, 降低数据源单一造成的事件误报率; (2)船舶行为动态建模技术: 利用知识图谱等技术融合船舶航行情境信息, 在不同船舶运动环境下利用深度学习、语义关联、图神经网络等方法构建不同的船舶行为模型, 提高检测准确性; (3)实时分析和可视化技术: 结合平行系统进行虚实系统间信息传递, 定性分析检测结果, 实时显示检测全过程, 提升监管过程中的人机交互效率。然后, 提出了包括数据采集、后台服务和客户端应用3个功能模块的交通事件平行检测系统; 该系统具备实时接收并处理船舶航行数据、分析并预测交通状态、动态检测并预警交通事件和仿真结果展示等功能。从数据融合、交通状态感知和交通虚实映射3个方面, 展望了面向海事监测实务的实时检测技术发展方向。      

单个交叉口模糊控制及Simulink仿真

以车辆排队长度为控制量,提出了一种交叉口模糊控制方法,并以四相位交叉口为例,建立了Simulink仿真模型,在不同交通流数据下将此模糊控制与现有的定时控制比较,对其控制效果进行仿真验证。仿真结果从车辆平均延误、信号周期和车辆的最大排队长度等方面,显示出了模糊控制的优越性。     

Effect of vehicle model on the estimation of lateral vehicle dynamics

A methodology is presented for estimating vehicle handling dynamics, which are important to control system design and safety measures. The methodology, which is based on an extended Kalman filter (EKF), makes it possible to estimate lateral vehicle states and tire forces on the basis of the results obtained from sinusoidal steering stroke tests that are widely used in the evaluation of vehicle and tire handling performances. This paper investigates the effect of vehicle-road system models on the estimation of lateral vehicle dynamics in the EKF. Various vehicle-road system models are considered in this study: vehicle models (2-DOF, 3-DOF, 4-DOF), tire models (linear, non-linear) and relaxation lengths. Handling tests are performed with a vehicle equipped with sensors that are widely used by vehicle and tire manufacturers for handling maneuvers. The test data are then used in the estimation of the EKF and identification of lateral tire model coefficients. The accuracy of the identified values is validated by comparing the RMS error between experimentally measured states and regenerated states simulated using the identified coefficients. The results show that the relaxation length of the tire model has a notable impact on the estimation of lateral vehicle dynamics.     

基于手机GPS定位数据的交通方式换乘点识别方法研究

交通方式换乘点识别长期以来是手机大数据交通调查领域的一大技术难点,既有研究大多通过设置出行时间、距离阈值进行识别,算法经验性强,普适性不佳,且易将起讫点、信号控制、交通拥堵等停留误识别为换乘停留。为此,提出了一种基于手机GPS定位数据的交通方式换乘点识别新方法：首先,构建模糊时空聚类算法识别个体运动-静止状态,算法同步实现了定位点时空密度双重聚类约束与聚类边界弹性需求,对个体运动状态识别效果更佳;其次,建立支持向量机模型进行交通方式换乘点识别,有效解决了起讫点、信号控制、交通拥堵等停留对换乘停留造成的干扰;最后,从出行链视角出发,提出了基于序列相似度算法的误差回溯自检与优化模型,能够有效修复换乘点漏识别与错误识别问题。此外,在成都市开展了大范围实测试验,由150名志愿者采集了近2 160 h得到的777.6万条数据被用于技术实证评估。试验结果表明：所述方法对交通方式换乘点平均识别准确率达89.3%,换乘时间平均识别误差控制在20 s以内;与既有空间聚类、小波分析算法相比,换乘点识别精度提升近10%,换乘时间误差最大可降低20 s以上,算法适用性与效果更佳。研究成果可为基于活动的交通需求模型演进提供数据支撑,为交通规划与管理部门决策提供技术支持。     

基于人体姿态估计的分心驾驶行为检测

为弥补现有驾驶特征提取方法的不足,提高分心驾驶行为检测的准确性和鲁棒性,将2D/3D人体姿态估计应用于驾驶人行为检测,提出一种适用于驾驶舱环境下的驾驶特征提取方法。首先通过将2D姿态估计网络Simple Baseline和分类网络ResNet进行融合,构建基于2D姿态估计的分心驾驶行为检测模型,并在分心驾驶数据集State Farm上分析不同数据增强方法、不同超参数、不同分类网络对模型性能的影响。其次,融合3D密集姿态估计网络DensePose与分类网络ResNet,构建基于3D姿态估计的分心驾驶行为检测模型。接着,在State Farm数据集上,针对模型的实时性和泛化能力,对比分析基于原始图像和基于2D/3D姿态的分心驾驶行为检测模型。最后,针对效果更优的基于2D姿态估计的分心驾驶行为检测模型,在分心驾驶数据集State Farm上,对使用不同姿态估计算法和分类网络的分心驾驶行为检测模型做了交叉试验,对比分析4个不同检测模型的优缺点。进一步地,将基于2D姿态估计的分心驾驶行为检测模型应用于实际采集的驾驶图片,对模型的泛化能力和有效性进行了测试验证。研究结果表明:与基于原始图像的检测模型相比,基于2D和3D姿态的检测模型都能显著提高分心驾驶行为的检测准确率;基于3D姿态的检测模型在检测精度方面略优,但基于2D姿态的检测实时性更好,检测效率是基于3D姿态检测的4倍;在驾驶舱单一环境下,基于2D姿态估计的分心驾驶行为检测模型能够满足分心驾驶行为检测的需求,在分心驾驶行为检测方面具有重要应用价值。