车联网人类动力学研究

发布时间：2018-04-13 10:37

本文选题：人类动力学 + 车载自组网　；参考：《吉林大学》2016年博士论文

【摘要】：人类动力学在互联网中的广泛存在现象已经被越来越多的学者研究和证实,各种网络系统中的非泊松现象被挖掘分析,基于动力学的网络模型也逐渐被建立。随着移动网络尤其是继智能手机之后的车载移动通讯的出现和快速发展,使得智能车联网络成为了新的网络研究热点,并且在其他网络系统中所发现的动力学因素是否同样存在并且作用于车联网中,以及以一种什么样的形式存在且影响着未来道路交通的状况值得关注。本文致力于挖掘车联网中的动力学因素,利用人类动力学知识和方法优化和建立智能车辆网络中的相关问题模型。重点分析了车辆可通讯的未来道路与传统道路的特性和不同,利用动力学方法对通讯方式进行了优化,并且根据人类动力学知识建立了新的无人驾驶车辆驾驶员的模型。主要贡献包括:1.分析了车辆网络中社会属性因素。通过对车辆网络中的相关概念进行定义,对三组公开车辆行驶信息数据分别进行建模、比较,通过社会网络的方法,分析车联网络中的社会行为属性。2.建立了智能驾驶和传统驾驶的车辆移动模型,比较了二者产生的道路拓扑的网络属性。根据智能驾驶的广视野性、预先判断和提前规划的特性,建立了基于智能驾驶和传统驾驶的两种车辆移动模型;进一步对两种移动模型进行同样道路的仿真实验、对所产生的道路拓扑进行了对比和分析,试图发现两种道路拓扑的不同,并且通过移动模型的建立找出道路拓扑网络属性不同的原因。3.提出了两种基于动力学的车联网通讯优化方法。(1)基于动力学的分簇方法:根据驾驶人的社会关系预先产生基于社会关系的驾驶员分组,进一步在行驶过程中,结合实时的位置和速度关系,产生车辆通讯的新的分簇方法;(2)基于动力学的跨层联合优化算法:在车辆通讯过程中,根据不同的利益要求,在引入容错机制后,根据观测动态地调节不同网络层的参数设置,并且通过群体博弈方法使得道路整体利益得到最大化,而不是使得车辆节点个体利益得到最大化。4.建立了基于人类动力学的驾驶员模型。通过分析影响驾驶员驾驶行为的动力因素,对该过程中的人类行为因素进行分类提取,计算出驾驶员动力学因子,进一步使该行为因子作用于驾驶过程中的四个子过程(这里,我们首次把驾驶过程分解为四个独立的动作,驾驶过程被认为是这四个动作的联合发生或者单独执行的结果),从而建立基于人类动力学的驾驶员模型。
[Abstract]:The widespread phenomenon of human dynamics in the Internet has been studied and verified by more and more scholars. The non-Poisson phenomenon in various network systems has been excavated and analyzed, and the dynamics based network model has been gradually established.With the emergence and rapid development of mobile network, especially after the smart phone, the smart car network has become a new research hotspot.And whether the dynamic factors found in other network systems also exist and act on the vehicle network, and what kind of form exist and affect the future road traffic situation is worthy of attention.This paper is devoted to excavating the dynamic factors in vehicle networking, optimizing and establishing the relevant problem models in intelligent vehicle network by using human dynamics knowledge and methods.In this paper, the characteristics and differences between the future road and the traditional road are analyzed, and the communication mode is optimized by means of dynamic method, and a new model of driverless vehicle driver is established according to the knowledge of human dynamics.The main contributions include: 1.The social attribute factors in vehicle network are analyzed.Through the definition of the related concepts in the vehicle network, three groups of open vehicle driving information data are modeled and compared, and the social behavior attributes of the vehicular network are analyzed by the method of social network.The vehicle moving models of intelligent driving and traditional driving are established, and the network properties of the road topology generated by the two models are compared.According to the characteristics of wide vision, prejudgment and advance planning of intelligent driving, two kinds of vehicle moving models based on intelligent driving and traditional driving are established, and the simulation experiments of the two kinds of moving models are carried out on the same road.By comparing and analyzing the road topology, we try to find out the difference between the two kinds of road topology, and find out the reason why the road topology network attribute is different by establishing the moving model.This paper proposes two dynamic optimization methods for vehicle networking communication. (1) A dynamic clustering method is proposed. According to the driver's social relations, the driver groups based on social relations are generated in advance, and further in the driving process,Combined with the real-time position and speed relationship, a new clustering method for vehicle communication is proposed. A dynamic cross-layer joint optimization algorithm is proposed: in the process of vehicle communication, according to different interests, after introducing fault-tolerant mechanism,According to the observation, the parameters of different network layers are dynamically adjusted, and the overall interests of the road are maximized by the group game method, rather than the individual interests of the vehicle nodes maximized. 4.The driver model based on human dynamics is established.By analyzing the dynamic factors that affect driver's driving behavior, the human behavior factors in the process are classified and extracted, and the driver's dynamic factors are calculated, which further make the behavioral factors act on the four sub-processes in the driving process.
【学位授予单位】：吉林大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：U495;TN929.5

【参考文献】