Fleet routing in transportation networks
Date of Issue2016-05-17
School of Electrical and Electronic Engineering
This project was undertaken by a group of three students. The purpose of this project was to analyse the taxi movement pattern and find taxi demand in order to optimize the use of taxis in Singapore. Before analyzing the taxi moving pattern, different algorithms should be researched in order to have a better understanding about how taxis operate. Some algorithms such as Genetic Programming, Tabu Search, Feasible Mobility on Demand (FMOD) and Real-time taxi dispatch Algorithms (Nearest Vehicle Dispatch and Insertion Heuristic) were researched. To analyse the taxi moving pattern and taxi demand, a huge confidential database was provided by SMART. One month data was provided on 2010 August, with an average 15000 taxis data in one day and 15000 times GPS tracking information for one taxi. Based on the assumptions on similar taxi demand at office hours on weekdays, weekends and public holidays respectively, whole month’s data was divided into these three parts. And the method used to analyze the data was, choosing one week data first, and then increasing to one month. Upon plotting graphs and making movies, some common patterns were found. But the data was chosen by only two days, and the results were accidental. Therefore, Joint and Individual Variation Explained (JIVE) method was applied. Scientific results were generated to estimate taxi travelling patterns and passenger travelling preference on weekdays and weekends. Matlab should be used as main tool to generate useful data and figures for analysis of taxi GPS data. Some movies were made to show the changes of taxis pick-up points within one hour in 5 minutes interval. Some 3D graphs were made to show the highest demand of taxis within one week. The purpose of using these movies and graphs was to have a better vision about the real situation.
Final Year Project (FYP)
Nanyang Technological University