The
Scottish
Branch

Evening lecture
Janny M.Y. Leung
Systems Engineering and Engineering
Management Department,
The
Wednesday
2nd April 2008, 6.30 p.m.
G38:
We propose a mixed
integer programming (MIP) optimization model for a timetable synchronization
problem for rail mass transit. The objective is to minimize the sum of all
waiting times of all passengers at interchange stations. By adjusting the
trains’ run-times and station dwell-times during their trips, and their
dispatch times, turnaround times and headways at the terminals, we can
construct high-quality timetables. A novelty in our formulation is the use of
binary variables to determine the relative sequencing of trains on different
lines with passenger transfers, which enables the correct representation of the
waiting times for transfers to the “next available” train at interchange
stations.
With trains departing
every few minutes from each terminal, there are a large number of trips to
consider, and hence the MIP formulation for the timetable synchronization
contains thousands of binary variables and tens of thousands of continuous
variables and constraints. We also investigated an optimization-based heuristic
for this problem, whereby we can get good-quality solutions within a reasonably
short time.
The train schedule for
the MTR system in