基于多源数据的通勤高峰期交通流量预测方法

发布时间：2019-02-19 12:57

【摘要】：随着中国城镇化进程加快,交通拥堵日益成为城市发展的通病,通勤高峰期尤为明显,通勤高峰期交通流量的精准预测是缓解交通拥堵的关键。另一方面,伴随着智能交通的发展和多学科的交叉融合,积累了不同类型的海量多源数据,如出租车GPS定位数据、人口户籍数据、参保、社保数据等,这些数据一定程度反映了通勤者的出行规律,为通勤高峰期交通流量的预测提供了数据支持。然而,如何从海量多源数据中分析时空关联关系,并在此基础上预测通勤高峰期的交通流量,还缺乏相应的解决方案。本文提出基于多源数据的通勤高峰期交通流量预测方法,依据不同交通工具的服务半径和出行距离,根据相关分类算法归纳总结通勤方式的最佳方案,为宏观交通规划的优化设计提供科学的决策。本论文研究主要内容包括:1.基于阐述本文的研究背景、研究意义、国内外交通流量预测相关进展,提出本文的研究内容;2.针对交通流量采集传感器进行归类整理,然后结合重庆市南岸区部分典型道路的真实断面交通流量数据,归纳总结出城市道路交通流量三类特征:动态性、时间相似性与空间相关性;3.多源数据预处理:从基本人口数据、路网数据、公交线路数据、出租车GPS轨迹数据等多源数据分析着手,提出了多源数据预处理方法,为交通流量预测提供数据准备;4.通勤高峰期交通流量预测原理:介绍利用经典Dijkstra算法实现最短通勤距离计算的非机动车出行统计模型;介绍基于轨道交通优先算法实现的私家车流量预测模型,.介绍利用地图匹配算法实现的出租车流量预测模型,进而实现基于多源数据的通勤高峰期交通流量预测研究。为验证本文方法的正确性,本文以武汉市为实验区域,通过实验证明,基于多源数据的通勤高峰期交通流量预测突破了依赖历史交通数据的局限,为交通流量预测提供了新的方法,为中微观交通规划和仿真提供新思路。
[Abstract]:With the acceleration of urbanization in China, traffic congestion is becoming a common problem in urban development, especially during the peak period of commuting. Accurate prediction of traffic flow during commuting peak is the key to alleviate traffic congestion. On the other hand, with the development of intelligent transportation and the cross-integration of multi-disciplines, it has accumulated different types of massive multi-source data, such as taxi GPS location data, population registration data, insurance, social security data, and so on. These data reflect the commuter's travel law to some extent and provide data support for the forecast of commuter's rush hour traffic flow. However, how to analyze the temporal and spatial correlation from massive multi-source data and predict the traffic flow during commuting rush hours on the basis of this, there is still a lack of corresponding solutions. In this paper, a multi-source data based commuting peak traffic flow prediction method is proposed. According to the service radius and travel distance of different vehicles, the best scheme of commuting mode is summarized according to the relevant classification algorithm. It provides scientific decision for optimization design of macro traffic planning. The main contents of this thesis are as follows: 1. Based on the background of this paper, the significance of the research, traffic flow prediction at home and abroad related progress, put forward the research content of this paper; 2. According to the traffic flow collecting sensors, this paper classifies and arranges the traffic flow data of some typical roads in Nanan district of Chongqing, and summarizes three characteristics of urban road traffic flow: dynamic, and dynamic. Temporal similarity and spatial correlation; 3. Multi-source data preprocessing: from the basic population data, road network data, bus line data, taxi GPS track data and other multi-source data analysis, put forward a multi-source data preprocessing method to provide data preparation for traffic flow forecasting; 4. The principle of commuting peak traffic flow prediction: this paper introduces a non-motor vehicle travel statistical model which uses classical Dijkstra algorithm to calculate the shortest commuting distance, and a private car flow prediction model based on rail transit priority algorithm. This paper introduces a taxi flow forecasting model based on map matching algorithm, and then realizes the research of commuting peak traffic flow prediction based on multi-source data. In order to verify the correctness of this method, this paper takes Wuhan as the experimental area and proves that the commuting peak traffic flow prediction based on multi-source data breaks through the limitation of relying on historical traffic data. It provides a new method for traffic flow prediction and a new idea for traffic planning and simulation.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U491.14

【相似文献】