公交车辆到站时间预测方法研究

发布时间：2018-12-17 00:00

【摘要】：公交车辆到站时间是公交智能化的重要体现之一,提高公交车辆到站时间预测的精度对于提升公共交通服务水平、缓解交通拥堵、减少乘客出行成本、实现公交系统的信息化具有重要的推动作用。首先,论文分析了公交运营数据采集的原理、方法和特点,设计了 GPS数据插值算法和公交线路离散化算法对公交运营数据进行处理。通过对公交车辆运行过程和到站时间的影响因素进行分析,把站点停靠时间和区间平均行驶速度作为输入变量,并设计了获取输入变量的算法。其次,论文以站点停靠时间和区间平均行驶速度为输入变量,建立了基于区间长度的统计方法到站时间预测模型(Statistical Method Bus Arrival Time Prediction Model Based on Interval Length,SMBATP-IL)、基于区间长度的卡尔曼滤波到站时间预测模型(Kalman.Filter Bus Arrival Time Prediction Model Based on Interval Length,KFBATP-IL)和基于区间长度的粒子滤波到站时间预测模型(Particle Filter Bus Arrival Time Prediction Model Based on Interval Length,PFBATP-IL),并设计了实现算法的具体流程和步骤。以PFBATP-IL模型作为算例,证明本文提出的预测方法和模型的可行性。最后论文选取北京市2条公交线路进行实证分析,以平均绝对误差MAE作为衡量预测结果的指标,选取早高峰(8:00)、平峰(11:00)和晚高峰(17:00)三个时刻,在不同的区间长度(10m,20m,30m)条件下,使用本文建立的模型进行到站时间预测。结果表明不同的区间长度对PFBATP-IL模型的影响最小,KFBATP-IL模型次之,对SMBATP-IL模型的预测结果影响最大。在最优的区间长度条件下,PFBATP-IL模型的预测结果是最优的,相比于KFBATP-IL模型和SMBATP-IL模型的预测结果分别改善了 16.86%和28.46%。
[Abstract]:The arrival time of public transport vehicles is one of the important embodiment of bus intelligence. Improving the precision of bus arrival time prediction can improve the level of public transport services, ease traffic congestion, and reduce passenger travel costs. Realizing the informationization of public transportation system has an important role to promote. Firstly, the paper analyzes the principle, method and characteristics of bus operation data collection, and designs GPS data interpolation algorithm and bus line discretization algorithm to deal with bus operation data. Based on the analysis of the factors affecting the running process and arrival time of public transport vehicles, the station stop time and the average driving speed of the interval are taken as input variables, and the algorithm to obtain the input variables is designed. Secondly, the paper takes the stop time and the average speed of the interval as input variables, and establishes the arrival time prediction model based on interval length (Statistical Method Bus Arrival Time Prediction Model Based on Interval Length,SMBATP-IL). Based on interval length Kalman filter arrival time prediction model (Kalman.Filter Bus Arrival Time Prediction Model Based on Interval Length,KFBATP-IL) and particle filter arrival time prediction model based on interval length (Particle Filter Bus Arrival Time Prediction Model Based on Interval Length,PFBATP-IL). The specific flow and steps of the algorithm are designed. The PFBATP-IL model is taken as an example to prove the feasibility of the proposed prediction method and model. Finally, two bus routes in Beijing are selected for empirical analysis. The average absolute error (MAE) is taken as the index to measure the predicted results. The three times of early peak (8:00), Pingfeng (11:00) and late peak (17:00) are selected. Under the conditions of different interval length (10m ~ 20m ~ 30m), the arrival time is predicted by using the model established in this paper. The results show that different interval lengths have the least influence on PFBATP-IL model, followed by KFBATP-IL model, and have the greatest influence on the prediction results of SMBATP-IL model. Under the condition of optimal interval length, the prediction results of PFBATP-IL model are optimal, which are improved by 16.86% and 28.46%, respectively, compared with those of KFBATP-IL model and SMBATP-IL model.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U491.17

【参考文献】