城市环境VANETs数据传输及智能安全行驶研究

发布时间：2018-07-31 18:06

【摘要】：随着汽车工业技术的快速发展,汽车的保有数量与日俱增,随之而来交通事故发生频率也迅猛增加。据统计全世界每天因交通事故死亡人数为3400人,每年受伤或致残人数数千万,直接经济损失5000亿美元。在现代汽车制造技术条件下,汽车本身设计造成交通事故所占比例越来越小,而驾驶员的个人因素成为造成交通事故的主要原因。为此,智能交通系统(Intelligent Traffic System)应运而生,它旨在将先进的信息技术、数据通信技术、传感器技术以及计算机技术等有效地综合运用于整个交通系统,加强交通系统中人——车——路之间的联系,从而减少交通事故的发生。与被动的安全防御技术相比,智能交通系统能主动发现威胁道路安全的危险源,并给司机予以警示,使其避免交通事故发生。因此智能交通系统更符合未来的发展需求,开展对其的研究具有重要的理论研究意义和工程应用价值。智能交通系统实现的关键技术为无线通信技术及智能感知技术,而车载自组织网络(VANETs)和智能手机感知作为两项技术的典型代表已引起了工业界和学术界的广泛关注,成为了当今研究的热点。本文在对VANETs进行了系统、全面地分析和总结的基础上,深入细致地研究了VANETs中数据传输问题。此外,本文还全面地分析了智能手机感知的特点,深入研究、探索其在智能交通系统中应用价值。针对这些问题,本文取得了若干创新和成果。本文的主要贡献包括:1.分析城市环境下VANETs数据传输模式以及影响网络连通性的因素,提出了车载自组织网络交通感知的数据传输策略TADS(Traffic-Aware data Delivery Scheme)。在该策略中,对于直路模式下数据传输,提出了改进的地理贪婪路由算法;对于路口模式下数据传输,通过计算候选道路效用函数,然后选择效用函数值最大的道路作为转发路径。在TADS中,效用函数由三部分组成,分别为道路车辆密度、车辆空间分布方差以及候选路口与目的地欧拉距离。车辆密度及车辆空间分布方差决定了链路的连通度,候选路口与目的地欧拉距离决定了数据的传输方向,道路效用函数值越大说明其链路连通性越好、候选路口离目的地越近。此外,为了减少由信息收集造成的网络开销,TADS根据车辆移动特性给出了交通流预测模型,大大减小了信息收集周期。通过实验仿真表明,TADS能以较小的网络传输代价及平均传输延迟完成较高的数据传输成功率。2.针对VANETs中大文件的数据分发,如图片、视频等,提出了车载自组织网络停车簇合作的数据分发机制COUP(COllaborative Urban Parking)。它主要利用城市环境下大量的停车簇作为天然的基础设施维持要分发的数据,通过停车簇之间的合作,为移动车辆提供数据分发服务。具体而言,对于每个内容下载请求,停车簇簇头首先按照先来先服务的原则,估计在本地为请求者提供的数据分发量。对于不能被分发的部分,簇头根据请求者的历史行程记录,建立其行程预测模型,进而选择出剩余内容被转发的目的停车簇,使请求者经过该停车簇可以继续未完成的内容下载。本文通过理论分析验证了COUP的有效性;基于真实交通数据的实验仿真结果表明,与另外两种数据分发算法相比,COUP有较高的成功下载率,尤其是在稀疏交通量和多下载请求的情况下。3.针对司机不安全驾驶行为严重威胁交通安全这一事实,提出了基于智能手机自矫正危险驾驶行为识别系统Driving Sense。它通过手机传感器感知车辆运动参数,监测车辆的行驶状态,能够识别超速、不规则改变行驶方向和异常速度控制三种危险驾驶行为。为了达到这一目的,首先从理论上分析了手机传感器噪声对车辆行驶状态估算的影响;然后,给出了运动状态下手机传感器误差分布确定方法,进而提出了基于改进卡尔曼滤波的手机传感器误差矫正算法,从而获得能更准确反映车辆行驶状态的数据。最后,运用矫正后的数据提出车辆的行驶状况估算方法。具体而言,为了估算车辆行驶速度,根据运动学知识提出了一种新颖速度估算方法;为了识别不规则改变行驶方向提出车辆转弯信号灯两阶段检测方法:(1)通过分析信号灯声音频率给出噪声过滤法;(2)对于去噪后的音频运用声音互相关操作检测转弯信号灯法;为了识别异常速度控制提出了阀值检测法。基于真实环境下实验结果表明,Driving Sense能有效识别出不安全驾驶行为。4.针对夜间司机视力受限交通事故频繁这一事实,提出了基于智能手机夜间司机辅助系统DNAS(Driver Nighttime Assistance System)。它主要利用智能手机摄像头感知主车辆后方车辆行驶情况,监测超速行驶或者近距离跟车的危险车辆,并给司机予以警示,使其获得更多反应时间应对危险突发事件。针对系统实现过程中的特有问题,提供了相应的解决方案。具体而言,首先给出了摄像头感知距离确定的方法;其次,基于车辆明亮大头灯以及它们之间的几何距离特性提出了夜间道路图片车辆识别算法;另外,根据车辆行驶的时空特性提出了图片中车辆跟踪算法,从而可以估算监测车辆行驶速度,最后,根据物体成像原理,提出了相对距离估算方法。基于真实环境下实验结果表明,在夜间DNAS能够有效监测超速行驶及近距离跟车事件。
[Abstract]:With the rapid development of automobile industry technology, the number of cars retained is increasing, and the frequency of traffic accidents is increasing rapidly. According to the statistics, the number of deaths from traffic accidents in the world is 3400 people every day, the number of injured or disabled people is tens of millions every year, and the direct economic loss is 500 billion dollars. The car itself is designed to cause more and more small proportion of traffic accidents, and the driver's personal factors have become the main cause of traffic accidents. Therefore, Intelligent Traffic System (Intelligent Transportation System) has emerged as the times require. It aims to integrate advanced information technology, data communication technology, sensor technology and computer technology. It is applied to the whole traffic system to strengthen the connection between people and vehicles in the traffic system, thus reducing the occurrence of traffic accidents. Compared with the passive safety defense technology, the intelligent transportation system can actively find dangerous sources of threat to road safety, and give warning to drivers to avoid traffic accidents. It is more in line with the future development needs, and it has important theoretical research significance and engineering application value to carry out its research. The key technologies for the realization of intelligent transportation system are wireless communication technology and intelligent perception technology, and the vehicle auto organization network (VANETs) and smart phone perception as the typical representatives of the two technologies have caused industrial circles. In this paper, based on the systematic analysis and summary of VANETs, this paper makes a thorough and detailed study of the data transmission in VANETs. In addition, this paper also analyzes the characteristics of the smart phone perception, deeply studies and explores its application price in the intelligent transportation system. A number of innovations and achievements have been made in this paper. The main contributions of this paper are as follows: 1. analysis of VANETs data transmission mode and the factors affecting network connectivity in urban environment, and propose a data transmission strategy TADS (Traffic-Aware data Delivery Scheme) for vehicle self-organizing network traffic perception (Traffic-Aware data Delivery Scheme). In the direct mode data transmission, an improved geographic greedy routing algorithm is proposed. For data transmission in the intersection mode, the road utility function is calculated by calculating the candidate road utility function, and then the path of the maximum utility function is chosen as the forwarding path. In TADS, the utility function is composed of three parts: road vehicle density and vehicle spatial distribution variance, respectively. And the distance between the candidate intersection and the destination Euler. The vehicle density and the variance of the vehicle spatial distribution determine the link connectivity. The distance between the candidate intersection and the destination Euler distance determines the transmission direction of the data. The greater the utility function of the road, the better the link connectivity is, the closer the candidate intersection is to the destination. In addition, in order to reduce the information collection. On the basis of the network overhead caused by the set, TADS gives a traffic flow prediction model based on the vehicle movement characteristics, which greatly reduces the information collection cycle. Through the experimental simulation, it is shown that TADS can accomplish high data transmission success rate with smaller network transmission cost and average transmission delay,.2. for data distribution of large files in VANETs, such as pictures, video and so on. The data distribution mechanism COUP (COllaborative Urban Parking) of car auto organization network parking cluster cooperation is proposed. It mainly uses a large number of parking clusters in the urban environment to maintain the data to be distributed as natural infrastructure, and provides data distribution services for mobile vehicles through the cooperation between parking clusters. For the content downloading request, the parking cluster heads first estimate the amount of data distribution provided locally for the requestor. For the part that can not be distributed, the cluster head establishes its travel prediction model according to the historical travel record of the requester, and then selects the destination cluster for the remaining content to be forwarded, so that the requester passes through. This parking cluster can continue unfinished content downloading. This paper validates the effectiveness of COUP through theoretical analysis. Experimental simulation results based on real traffic data show that COUP has a higher successful download rate compared with the other two data distribution algorithms, especially in the case of sparse traffic and multiple download requests,.3. is not for drivers. The fact that safe driving behavior seriously threatens traffic safety, proposes a Driving Sense. based on smart phone self correcting dangerous driving behavior identification system, which can perceive vehicle motion parameters through mobile phone sensors, monitor the driving state of vehicles, identify overspeed, irregular change driving direction and abnormal speed control three kinds of dangerous driving In order to achieve this goal, the influence of cell phone sensor noise on vehicle driving state estimation is analyzed theoretically. Then, the method of determining the distribution of cell phone sensor error in motion state is given, and then the error correction algorithm based on improved Calman filter is proposed, which can be more accurately reflected. In the end, the method of estimating the driving condition of the vehicle is put forward by using the corrected data. In order to estimate the speed of the vehicle, a novel speed estimation method is put forward according to the kinematics knowledge. In order to identify the irregular change direction, the two phase detection method of the vehicle turning signal lamp is proposed: (1) The noise filtering method is given for the over analysis of the sound frequency of the signal lamp; (2) the turn signal method is used to detect the noise after the noise reduction. In order to identify the abnormal speed control, the threshold detection method is proposed. Based on the real environment, the experimental results show that the Driving Sense can effectively identify the unsafe driving behavior.4. for night drivers. The fact that vision limited traffic accidents are frequent is based on the DNAS (Driver Nighttime Assistance System) based on the smart phone night driver. It mainly uses a smart phone camera to perceive the driving condition of the rear vehicles in the rear of the main vehicle, monitors the dangerous vehicles running over speed or near the car, and gives the driver a warning to the driver. More response time should be obtained to deal with dangerous emergencies. In view of the specific problems in the system implementation, the corresponding solutions are provided. In particular, the method of determining the perception distance of the camera is given first. Secondly, the night road picture vehicle recognition is proposed based on the bright headlights and the distance between them. In addition, the vehicle tracking algorithm in the picture is proposed based on the space-time characteristics of the vehicle, and the vehicle speed can be estimated. Finally, the relative distance estimation method is proposed based on the principle of object imaging. Based on the real environment, the experimental results show that at night DNAS can effectively monitor the speeding and near distance car following. Event.
【学位授予单位】：电子科技大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：U495

【相似文献】