城市高架道路交通实时仿真及预测

交通仿真是交通领域的一个研究热点。通过真实的模拟行人、车辆、交通设施的运行状况,以对某个路口、某片区域的交通情况进行仿真。根据仿真得到的数据进行方案的评价。或者通过快速仿真,基于现实的交通状况和历史数据,对未来的交通情况进行预测。本文主要论述了如何对城市高架道路进行实时再现仿真,并对预测提出了一定的设想。     

Real-time road traffic states measurement based on Kernel-KNN matching of regional traffic attractors

The accurate measurement of road traffic states can provide decision making for travelers and traffic managers. In this paper, an algorithm based on Kernel k-nearest neighbors (Kernel-KNN) matching of regional traffic attractors is presented to estimate road traffic states. First, the road traffic running states are divided into several different modes. The concept of the regional traffic attractors of the target link is put forward for effective matching. The representative road traffic state data are extracted to establish the reference sequences of road traffic running characteristics (RSRTRC). Then the sequence of regional traffic attractors is selected and its kernel function is constructed, with which the regional traffic attractors can be mapped into a high dimensional feature space. The reference and current sequences of regional traffic attractors are extracted and the Euclidean distances in the feature space between them are obtained. Finally, the road traffic states are estimated from weighted averages of the selected k road traffic states, which correspond to the k smallest Euclidean distances. Several typical links in Beijing are adopted for case studies. The final results of the experiments are presented, which prove that this road traffic state measurement approach based on Kernel-KNN matching of regional traffic attractors is feasible and can achieve a high accuracy.     

Study of a simulator database and an experiment for the evaluation of highway driving safety

Drivers acquire more than 90% of visual information through road signs. However, the road signage along highways can cause driver discomfort and traffic accidents. Conducting experiments along a real highway may cause traffic accidents; therefore, obtaining accurate results from such experiments is difficult. Consequently, this paper reports an experiment for evaluating highway driving safety to reduce the causes of accidents through highway design. It examines the development of a highway database to classify road types in exact accordance with real highways. The study involves an experiment using a driving simulator. The highway database was constructed in two aspects (original and alternative designs) compatible with the DirectX program of Multigen. Four types of physiological measurements (i.e., electrocardiogram, electromyography, electrooculography, and galvanic skin response) were taken from 33 subjects using Biopac equipment in the developed highway database. The skin temperature was measured by a thermal camera. The driver’s state was measured through the Subjective Workload Analysis Technique, NASA-Task Load Index, and Modified Cooper-Harper, among various psychological methods. Statistical analysis was conducted to assure the reliability of the results using the Statistical Package for the Social Sciences 12.0 program.     

基于LSTM-RF模型的城市道路交通流预测

交通流预测作为信号协调和出行时间预测等任务的基础,成为了交通领域的研究点。对于交通流预测问题,研究人员提出了多种方法,但这些方法大多只使用交通流数据的时域信息进行交通流预测,忽略了空间相关性对于预测目标路段流的影响,导致预测精度不理想。基于组合模型的思想提出了一种称为LSTM-RF的交通流预测模型。在交通流预测过程中,首先使用LSTM模型提取预测目标路段的时序特征,再将其预测值与采集的相邻上下游路段信息同时作为随机森林模型的输入特征,进行交通流时空相关性分析,获得最终的预测结果。并通过贵阳市车牌识别系统采集的城区132条路段的交通流数据进行实验验证。结果表明:该方法在预测精度上优于单一模型,并且预测误差相比单一模型有明显减少。     

基于CCD的待车数传感器研究--等待车辆图像的腐蚀-膨胀识别算法

城市交通系统是一个复杂系统，为实现城市汽车流实时响应控制，交通信息的实时采集和处理显得尤为重要。针对现有车辆检测器需对道路进行破坏性改造等问题，研究一种基于单幅、静态图像处理的CCD路口等待车辆数传感器；提出在对图像连行有效的预处理的基础上，采用腐蚀-膨胀识别算法，使系统实时检测出交通路口的等待车辆数，并以此作为城市交通信号控制系统实时决策的状态信息。通过示例证明了算法的实时性、健壮性及自动化的应用前景。     

高级驾驶辅助系统前视摄像头硬件在环测试

高级驾驶辅助系统可以大幅度降低交通事故,减少人员伤亡和经济损失,摄像头是该系统中的重要部件。为了在前视摄像头开发期间更早发现存在的问题,模拟路试过程中的危险工况,提出了一种基于CarMaker和VeriStand的硬件在环测试系统。应用所搭建的台架进行了大量摄像头的硬件在环功能测试,台架测试结果与实车测试结果一致,在车道偏离预警和前方碰撞预警典型工况下均能及时产生报警信号。     

超视界路况下网联汽车的辅助驾驶方法

针对车辆前方超视界路况不易预测、驾驶员易因路况不明而错误操作车辆导致失稳等危险事故的情况，提出了一种超视界路况下的网联汽车辅助驾驶方法。设计了基于车联网实时通信技术的网联汽车辅助驾驶系统，它主要由云端服务、共享驿站和车载单元组成；研究了行驶车辆的路况信息识别方法，包括路面附着系数识别、道路曲率计算和路面坡度测量；研究了基于多车辆测量数据的路况信息融合与校正方法，建立了基于三自由度车辆的刚体力学的弯道安全车速计算模型。实车实验结果表明，所提方法能够准确测量车辆前方超视界的路况信息，并可使驾驶员提前、准确地获知超视界路况信息，特别是弯道路况信息。     

Dual extended Kalman filter for combined estimation of vehicle state and road friction

Vehicle state and tire-road adhesion are of great use and importance to vehicle active safety control systems. However, it is always not easy to obtain the information with high accuracy and low expense. Recently, many estimation methods have been put forward to solve such problems, in which Kalman filter becomes one of the most popular techniques. Nevertheless, the use of complicated model always leads to poor real-time estimation while the role of road friction coefficient is often ignored. For the purpose of enhancing the real time performance of the algorithm and pursuing precise estimation of vehicle states, a model-based estimator is proposed to conduct combined estimation of vehicle states and road friction coefficients. The estimator is designed based on a three-DOF vehicle model coupled with the Highway Safety Research Institute(HSRI) tire model; the dual extended Kalman filter (DEKF) technique is employed, which can be regarded as two extended Kalman filters operating and communicating simultaneously. Effectiveness of the estimation is firstly examined by comparing the outputs of the estimator with the responses of the vehicle model in CarSim under three typical road adhesion conditions(high-friction, low-friction, and joint-friction). On this basis, driving simulator experiments are carried out to further investigate the practical application of the estimator. Numerical results from CarSim and driving simulator both demonstrate that the estimator designed is capable of estimating the vehicle states and road friction coefficient with reasonable accuracy. The DEKF-based estimator proposed provides the essential information for the vehicle active control system with low expense and decent precision, and offers the possibility of real car application in future.     

Measuring system for investigation of tri-axial stress distribution across the tyre-road contact patch

A measuring system for investigation of stress distributions in the tyre-road contact patch was designed and developed. The transducer within this system comprises a transversal array of sensing elements, covering the entire contact patch width in a single run. The system simultaneously measures stress distributions on three directions for a truck tyre (or even airplane tyre). Each sensing element has 10 mm × 10 mm contact surface and optimized dimensions, allowing measurements in various wheel rolling conditions. The transducer induces minimum changes in tyre-road contact properties, as it has very small gaps around sensing elements. The system is road mounted, in real rolling conditions. The measuring system contains 90 strain measuring channels. Main aspects regarding system calibration and specially developed software are illustrated. Preliminary results of tyre-road contact stress distributions are presented.     

基于小波去噪和LSTM模型的短时交通流预测

针对现有的部分交通流预测模型仅面向单一路段进行,模型输入数据未预处理的问题,采用启发式阈值算法对小波分解后的原始交通流数据进行去噪处理,通过对路网内各路段交通流数据相关性系数计算,构造出路网交通流数据压缩矩阵。数据去噪将数据对模型的干扰降到最低,同时路网数据相关性分析又使预测在路网层面上进行了考量。利用长短时记忆(LSTM)网络在时序数据处理方面的优势,将压缩矩阵输入构造好的LSTM模型进行短时交通流预测。利用去噪处理数据和原始数据分别训练LSTM-1和LSTM-2模型,通过仿真实验,设置不同预测时间将本文提出的预测方法和其他4种模型对比,验证了相较于其他4种模型预测的准确率平均可提升10.278%,预测的准确率达到了95.58%,说明这是一种高效率的短时交通流预测方法。     

基于软硬件协同设计的车道线实时识别方法

为兼顾车道识别的鲁棒性和实时性,综合考虑硬件设计资源和软件功能的有效分配,设计了一种基于嵌入式平台的车道线识别方法。从整个系统的生产消费模型考虑,提出了匹配增强的必要性,并根据车道线不同宽度特征,提出了多尺度匹配滤波方法;根据改进的恒虚警率检测数学模型来估计局部噪声水平从而确定最适应的动态提取阈值。以可编程逻辑门阵列(FPGA)采集高帧率、宽动态范围的道路图像,并以系数分解法实现匹配滤波的硬件加速优化设计。实验和评估结果证明,该嵌入式系统可以在多种交通场景下鲁棒地识别车道线特征点,处理速率达到每秒60帧。     

Stress-In-Motion (SIM) system for capturing tri-axial tyre–road interaction in the contact patch

A unique measuring system for the quantification of tri-axial (3-D) tyre contact force (or stress) distributions was designed, developed and used in several studies since 1994. The uniqueness of the system is defined by a textured measuring surface in order to represent a typical “textured” road surface. The system is referred to as the Stress-In-Motion (SIM) system. A single SIM measuring pad testing area comprises a total of 1020 supporting pins and a transverse array of 21 sensing elements, covering the entire tyre contact patch with in a single run. The SIM pad measuring system is modular in concept, allowing multiple SIM measuring pads to be used for e.g. wide base truck tyres, or a dual tyre configuration, or full axle measurements – using a quad SIM pad measuring system configuration. Tyre contact force (or stress) distributions are simultaneously captured in the three orthogonal directions X, Y and Z for a single, dual or full axle truck tyre configuration. Each of the sensing elements has a 9.7 mm diameter circular contact surface area (∼73.9 mm²) and is dimensionally optimised, allowing measurements in various tyre rolling conditions on a textured measuring surface. The textured surface induces some pre-conditioning of tyre–road contact properties, as it has small gaps around all supporting and measuring pins. The system is installed flush with the road surface, preferably on a rigid support base, and can be used for real tyre (or truck) rolling conditions. A single SIM measuring pad contains 63 strain measuring channels (3 × 21) for the sensing elements. Aspects such as SIM system design, sensing element calibration, system usage and outputs of specially developed software are illustrated. Several results of tri-axial road contact stress distributions are also presented.     

可行区域视觉感知的一种设置方法

提出了一种在ROI（Region Of Interest）即感兴趣区域中建立搜索窗口的新的车道识别算法，使其能够更好的处理非理想路况的各种不确定因素，并建立实时动态的搜索带，除了能更精确的追踪车道标志线之外，还能更好的应用于弯路路况。由于ROI的建立，整个识别过程的计算量大大降低。因此更适合实时要求。     

Engine oil contribution to diesel exhaust emissions

The reduction of harmful pollutant emissions as well as CO₂ emissions emanating from motor vehicles will be of considerable interest in the coming decades. Emissions legislation will be the guiding principle in the development of new technologies and vehicles. More attention will have to be paid to off‐road vehicles, especially tractors, if the production of healthy food and the maintenance of a cleaner environment are not to be compromised. Therefore, one the biggest challenges facing the automotive industry is to improve fuel economy, both to conserve natural resources and to limit pollutants and CO₂ emissions. Better fuel efficiency and consequently lower emissions will require new materials, new lubricants and low‐emission fuel. Engine lubricants help to improve vehicle efficiency but contribute engine exhaust emissions. This paper deals with the influence of engine lubricants on diesel exhaust emissions. Investigations have shown a clear effect of lubricant oil on emissions, which depends on lube oil characteristics, especially sulfur content, metal content, volatility and density. Copyright © 2006 John Wiley & Sons, Ltd.     

Road profile estimation using neural network algorithm

This paper more specifically focuses on the estimation of a road profile (i.e., along the “wheel track”). Road profile measurements have been performed to evaluate the ride quality of a newly constructed pavement, to monitor the condition of road networks in road management systems, as an input to vehicle dynamic systems, etc. The measurement may be conducted by a slow-moving apparatus directly measuring the elevation of the road or using a means that measures surface roughness at highway speeds by means of accelerometers coupled with high speed distance sensors, such as laser sensors or using a vehicle equipped with a response-type road roughness measuring system that indirectly indicate the user’s feelings of the ride quality. This paper proposes a solution to the road profile estimation using an artificial neural network (ANN) approach. The method incorporates an ANN which is trained using the data obtained from a validated vehicle model in the ADAMS software to approximate road profiles via the accelerations picked up from the vehicle. The study investigates the estimation capability of neural networks through comparison between some estimated and real road profiles in the form of actual road roughness and power spectral density.     

基于接触摩擦非线性特征的车-人交通事故数值再现分析

车-人接触碰撞事故作为一种典型的交通事故类型,包含车、人、地三个系统,涉及变量众多,再现过程复杂。其中,人-地接触摩擦因数作为典型的输入参数,对事故再现结果有着重要影响。传统的交通事故数值再现计算中,摩擦因数一般作为定值输入,其仿真结果与实际物理过程存在较大偏差。通过纤维-粗糙路面接触的简化物理模型探讨车-人交通事故中人-地接触摩擦的非线性关系,提出针对车-人碰撞交通事故的人-地摩擦因数与法向力函数关系的构造方法,并通过典型车-人碰撞事故案例对其进行了验证。案例分析表明,基于上述方法构造的摩擦因数数学模型可减小仿真结果与实际物理过程的偏差,从而为交通事故鉴定提供更为准确有力的技术依据。将接触摩擦力学应用于交通事故再现数值计算的思想为后续相关研究提供了参考。     

Hydrogen refuelling station calibration with a traceable gravimetric standard

Of all the alternatives to hydrocarbon fuels, hydrogen offers the greatest long-term potential to radically reduce the many problems inherent in fuel used for transportation. Hydrogen vehicles have zero tailpipe emissions and are very efficient. If the hydrogen is made from renewable sources, such as nuclear power or fossil sources with carbon emissions captured and sequestered, hydrogen use on a global scale would produce almost zero greenhouse gas emissions and greatly reduce air pollutant emissions.The aim of this work is to realise a traceability chain for hydrogen flow metering in the range typical for fuelling applications in a wide pressure range, with pressures up to 875 bar (for Hydrogen Refuelling Station - HRS with Nominal Working Pressure of 700 bar) and temperature changes from −40 °C (pre-cooling) to 85 °C (maximum allowed vehicle tank temperature) in accordance with the worldwide accepted standard SAE J2601. Several HRS have been tested in Europe (France, Netherlands and Germany) and the results show a good repeatability for all tests. This demonstrates that the testing equipment works well in real conditions. Depending on the installation configuration, some systematic errors have been detected and explained. Errors observed for Configuration 1 stations can be explained by pressure differences at the beginning and end of fueling, in the piping between the Coriolis Flow Meter (CFM) and the dispenser: the longer the distance, the bigger the errors. For Configuration 2, where this distance is very short, the error is negligible.     

Laser-triangulation device for in-line measurement of road texture at medium and high speed

The knowledge of the friction coefficient between road and tyres is a very precious information to implement vehicle active control, especially considering the optimisation of the braking action. One of the most important parameters able to influence such a coefficient is the road texture (microtexture and macrotexture). Although different methods are now available to perform reliable texture measurements, no well-established techniques currently exist for measuring the pavement texture at medium-high speed with real time data analysis during the usual vehicle operations.This paper presents a method, based on two identical industrial laser-triangulation displacement transducers, allowing to get real time reliable road micro and macrotexture measurements during standard vehicle operations, even at medium and high speed. The presence of two transducers also allows to estimate the instantaneous vehicle speed, which is needed to obtain the road texture from the sensor time-histories. This means that the presented system can be considered as a stand-alone device able to give as an output the road texture (micro and macro) and also the vehicle speed without any other input. The paper also underlines the advantages of this method and its drawbacks. The method reliability is evidenced by some real time outdoor tests on the different road surfaces of the Pirelli test track and at different vehicle speeds.     

基于Tchebichef不变矩与SVM的交通标志分类算法研究

为了更好地对交通标志进行分类,提出了一种两步分类算法。首先根据交通标志的颜色与形状特征设计多层决策树进行粗分类,然后采用径向Tchebichef不变矩提取交通标志的特征,并基于SVM设计分类器进行精分类。对从3000幅自然场景图像中检测出的交通标志进行分类实验,结果表明该分类系统能取得较高的分类精度,这为交通标志识别系统的实现奠定了基础。     

提高车道线识别精度的方法

提出了一种在前方车道图像中建立高分辨率的ROI(Region Of Interest)即感兴趣区域,并在其中建立搜索带以提高车道识别精度的方法,实验证明这种方法能够更好的处理非理想路况的各种不确定因素,除了能更精确的追踪车道标志线之外,还能更好的应用于前方图像出现噪声、阴影以及路面反光等不利情况.由于高分辨率ROI的建立,在不增加运算时间的前提下,识别精度却更加提高,同时整个运算过程的计算量大大减少.