AbstractWith published timetables in railway system, punctuality is an important factor which effects degree of customer satisfaction. Promoting the system Reliability (punctuality) is an effective and low-cost method without adding any infrastructure and staff. Therefore, it’s important to resolve delay problems for promoting punctuality in railway transportation. In the case of Taiwan railway administration (TRA), it is complicated to release the key factors in traditional railway system with multiple service type, single-double track and multiple types of platform. If the delay reasons and the interactions among the delay factors can be clearly clarified, the impacts of headway can be exactly handled. Furthermore, reliability and service quality of railway operation can be enhanced.
Due to the record characteristics of delay events in TRA, it is difficult to catch the key factors of delay reasons by historical data. The study adopts a supervised decision tree method in machine learning techniques, which is named C4.5, to estimate the key factors of delay. In this study, a delay root cause mining method is designed to discover the root cause delay factor by logic analyzing the trains waiting behavior which is caused by scheduled or un-scheduled meetings and overtaking. The delays can be resolved by the adjustment of timetable, and discovered by frequency filtering which would be an important reference for the next timetable rescheduling. The result of this study can be applied as a reference for the railway system, especially in timetable rescheduling, system reliability analysis and service quality improvements.
|Date of Award||2012|
|Supervisor||Wei-Hsun Lee (Supervisor)|