A Study on the Modeling of Delay and Travel Time for City Bus

• Main Authors: Kai-Bin Chen, 陳凱斌
• Other Authors: none
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/16013092068269251375

碩士 === 逢甲大學 === 交通工程與管理所 === 92 === In commuters` opinion, they have two times in their mind, one is when a bus is coming, i.e. arrival time, and the other is how long they will spend traveling to their destinations, i.e. travel time, for a regular city bus route. The times can be used to check whether a bus runs on time or not. Because there were few studies focused on this on time or not topic in the past, this study wants to know what the important affecting variables are. This study chooses two city bus routes; one in Taipei and the other in Taichung. It wants to prove that for passenger segments belonging to different times and different ways, the effects of variables on delay or travel time variation should be different. Because the data in Taichung were recorded for every service on a route and it there is no recorded data in Taipei, this study uses created data for Taipei. Created data in Taipei were used in the accelerated failure time model (one of survival models) to check if the models are good ones to fit the mentioned a data, especially the case of the estimated results of the delay travel time model. This study lays emphasis on the part of Taichung. The input data of models for Taichung were recorded by persons who have been on the buses of line “25” all day long for a week. According to both temporal and spatial passenger segmentation, the study separates the data set into four parts: weekday along one way, weekday along the other way (at the contrary direction), weekend along one way and weekend along the other way. This study says that there are supposed different parameters in different passenger segments. The other variables include “station sequences”,” numbers of bus stops at stations”,” numbers of stops waiting for red light”,” numbers of the situation when traffic lights turn from red to green and the bus is still not moving”, “numbers of the passengers going aboard and alight at bus stations”,” whether the bus operates during peak hours or not”,” whether the previous bus run returns late or not”. Because there are two ways to describe if a bus is on time or not, i.e. early or late, therefore early or late time models are used for both arrival and travel time. The study uses survival models to analyze the difference between a real and a predefined times both for arrival and travel, and uses logit models to know whether the bus running a certain route early or late. By summing up all possible results, there are sixteen survival models and eight logit models. Most significant variables from all travel time models are “station sequences”,” numbers of stops waiting for red light” and” numbers of the situation when traffic lights turn from red to green and the bus is still not moving”. The number of the significant variables of the models for the way ending at city center is greater than that starting from the city center, and the number of the significant variables of the models for weekday is greater than that for weekend. The variable,” whether the previous bus run returns late or not “has no remarkable effect on any travel time model. The variables of all delay time models such as “station sequences “and” numbers of stops waiting for red light” are all much remarkable. Contrary to the results of delay time models, those of travel time models indicate the number of the significant variables of the models for the way starting from city center is greater than that ending at the city center and the number of the significant variables of the models for weekend is greater than that for weekday. Finally, in concluding the similar results of both the travel time and delay time by logit models, the estimated parameters all mean late bus operations, so this phenomenon is quite common.