基于移动设备的路况估计算法研究

发布时间：2018-01-05 20:06

本文关键词：基于移动设备的路况估计算法研究　出处：《北京邮电大学》2014年硕士论文　论文类型：学位论文

【摘要】：道路交通拥堵是燃油利用率低,资源浪费的主要原因。准确估计道路交通状况、定位交通拥堵是减少人们出行时间的重要一步,也是智能交通系统中的重要课题之一。目前已有的路况估计方法主要有三种,分别是基于路边固定单元,基于无线探测与定位和基于车载GPS定位与通信。感应线圈,RFID射频卡等路边固定单元,车载定位与通信模块安装和维护费用高,覆盖率低,利用基站的无线探测与定位受移动网络影响严重,都不能很好地实现路况估计。随着智能手机的普及,利用手机集成的GPS定位作为交通探测器得到广泛关注,使用从司机的手机中获取的GPS信息监控交通,开辟了交通状况估计的新途径。论文将构建基于智能手机GPS定位功能的路况估计系统架构。首先,对单车LATT (Link Average Travel Time,路段行程时间)的估计算法进行探索性研究,使用三次埃米尔特插值算法,提高车辆通过路段端点时刻的估算准确性。其次,由于同类型车辆具有相似的运动特征,充分考虑单车对路段平均速度的贡献,采用权重法对单车路段平均速度进行融合得到同类车辆的路段平均速度。再次,不同状况下,路段平均速度与车辆的平均速度映射关系复杂,鉴于此,引入RBF (Radial Basis Function,径向基函数)网络,以同类型车辆路段平均速度和不同类型车辆数目的比值作为输入计算路段平均速度。最后,介绍了表征道路交通状况的重要参数,针对手机GPS数据获取量较小的情况,引入速度密度互推导算法构建新的路况估计模型。实验结果表明,三次埃米尔特插值算法、同类车辆平均行程速度融合算法、RBF神经网络法，，都降低了相应参数的估计误差,提高了系统对路段平均速度的估计准确性。此外,在低渗透率下,基于速度密度互推导算法构建的补充方案,对于降低路段平均速度和车辆密度的估计误差也有明显效果。
[Abstract]:Traffic congestion is the low utilization rate of fuel, mainly due to the waste of resources. The accurate estimation of road traffic conditions, location of traffic congestion is an important step in reducing people to travel time, but also in the intelligent transportation system is one of the important topics. The existing traffic estimation method mainly has three kinds, respectively is fixed roadside unit based on wireless detection with the positioning and GPS positioning based on vehicle and communication. Based on the induction coil, RFID radio frequency card fixed roadside unit, vehicle positioning and communication module installation and high maintenance cost, low coverage, using a wireless base station probe measurement and positioning by the mobile network influence is serious, can well realize the traffic estimation. With the popularity of smart mobile phone the use of mobile phone GPS positioning integrated as traffic detector to get attention, GPS information obtained from monitoring traffic drivers in the mobile phone, open up traffic. A new way to plan.
The construction of intelligent mobile phone GPS positioning function based on the traffic estimation system architecture. Firstly, the LATT (Link Average Travel Time bike, travel time) of an exploratory study of the estimation algorithm using three Hermite interpolation algorithm, improve the vehicle by estimating the accuracy of Lu Duanduan point. Secondly, due to the same type of vehicle with motion characteristics similarly, fully consider the bike section average velocity contribution was obtained by fusion of similar vehicles on the bike section average velocity of section average velocity using the weight method. Thirdly, under different conditions, the average velocity mapping section average velocity and vehicle complex relationship, in view of this, the introduction of RBF (Radial Basis Function, the radial basis function network) the ratio of the number of vehicles of the same type, average road speeds and different types of vehicle as the input calculation of section average velocity. Finally, the table In view of the important parameters of the road traffic conditions, in view of the small amount of GPS data acquisition of mobile phone, a new estimation model of road condition is constructed by introducing the speed density mutual deduction algorithm.
The experimental results show that the three Hermite interpolation algorithm, fusion algorithm of average travel speed of similar vehicles, RBF neural network method can reduce the estimation error of the corresponding parameters, improve the estimation accuracy of section average velocity. In addition, in low permeability, supplementing the construction speed of the algorithm is derived based on density, also has the obvious the effect for reducing the error estimates of section average velocity and density of vehicles.

【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U495

【参考文献】