基于位置和乘车信息的公交站点客流预测方法

发布时间：2018-03-23 06:09

本文选题：位置　切入点：乘车信息　出处：《山东大学》2017年硕士论文　论文类型：学位论文

【摘要】：在交通拥挤日趋严重、交通污染与交通事故等问题日益突出的背景下,优先发展公共交通是缓解大中城市交通问题最有效的手段之一。公共交通的发展同时可以促进城市经济的发展,公共交通已成为城市客运交通系统的主体,成为国家在基础建设领域中重点支持发展的产业之一。传统的公交客流统计方式存在调查步骤繁琐、成本高、精确度低等一种或多种弊端,本文通过对地面常规公交自动售票系统收集的数据进行数据挖掘,对刷卡信息进行统计分析,得到公交客流出行特性和规律,更加高效、精确地预测公交客流,并有效地利用了现有交通信息资源。本文在查阅相关研究文献的基础上,从站点客流估计、客流特性分析和交通短时预测方法三个方面对国内外研究成果进行综述,总结对比现有研究成果,指出本论文在此基础上的研究重点。首先,对公交自动售票系统进行数据特征分析,根据数据特征和客流估计需求对原始数据进行数据筛选和数据剔除预处理,对刷卡记录进行聚类分析,融合自动售票系统数据、GPS数据、静态路网和公交调度信息等多源数据,进行位置和乘车信息的时间匹配,得到匹配站点,估计站点客流。其次,基于多源数据融合的站点匹配算法,以济南市智能公共交通系统数据为例,从客流的方向不均衡性和供需平衡性对线路客流进行特性分析,从客流的通勤特性和时间不均衡性对站点客流进行特性分析,针对站点的通勤和时间特性选取线路的特征站点,为客流预测提供案例站点。最后,对影响站点客流的因素进行变量分析,选取时间序列模型和改进BP神经网络模型对特征站点进行客流预测,比较不同预测模型对不同通勤类型站点客流预测的精度,得出站点客流预测的结论。
[Abstract]:Against the background of increasingly serious traffic congestion, traffic pollution and traffic accidents, Giving priority to the development of public transport is one of the most effective means to alleviate the traffic problems in large and medium-sized cities. The development of public transport can also promote the development of urban economy, and public transport has become the main body of urban passenger transport system. It has become one of the industries that support the development of the country in the field of infrastructure construction. The traditional way of bus passenger flow statistics has one or more disadvantages, such as tedious investigation steps, high cost, low precision and so on. In this paper, the data collected by the ground bus automatic ticket selling system are mined, and the information of credit card is statistically analyzed, and the travel characteristics and rules of public transport flow are obtained, so as to predict the bus passenger flow more efficiently and accurately. On the basis of consulting relevant research literature, this paper summarizes the domestic and foreign research achievements from three aspects: station passenger flow estimation, passenger flow characteristic analysis and traffic short-term forecasting method. This paper summarizes and compares the existing research results, and points out the key points of this paper. Firstly, the paper analyzes the data characteristics of the bus automatic ticket system. According to the characteristics of data and the demand of passenger flow estimation, the original data are filtered and pre-processed, the credit card records are clustered and analyzed, and the GPS data, static road network and bus dispatch information are fused, and the multi-source data, such as GPS data, static road network and bus dispatch information, are fused. Matching the location with the time of the ride information, getting the matching station, estimating the passenger flow of the station. Secondly, the station matching algorithm based on multi-source data fusion, taking Jinan intelligent public transportation system data as an example, The characteristics of line passenger flow are analyzed from the disequilibrium of passenger flow direction and the balance of supply and demand, the characteristic analysis of station passenger flow is carried out from the characteristics of commuting and time imbalance of passenger flow, and the characteristic station of line is selected according to the commuting and time characteristics of the station. Finally, the factors influencing the passenger flow are analyzed, and the time series model and the improved BP neural network model are selected to forecast the passenger flow of the characteristic stations. This paper compares the accuracy of different forecasting models for passenger flow of different commuting stations, and draws the conclusion of forecasting passenger flow at different stations.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U491.17

【参考文献】