MTR :: Train Timetabling

Introduction

Producing optimized train schedules and timetables is highly complex. By combining a model of depot operations with control and operational rules and constraints, constraint-programming techniques can help improve customer satisfaction and ensure safe operations.

Using optimization techniques, MTRC is now able to provide better service by increasing the number of trains operating during peak hours.

Our Client

Mass Transit Railway Corporation (MTRC), established in 1975, is one of the most heavily utilized underground railway networks in the world. Since the opening of the first line in 1979, the MTR has been providing a safe, reliable, and efficient service to passengers. The current subway is now over 88 km with 6 lines through 50 stations with a new Disneyland Resort Line opening in late 2005.

Business Needs

In 1995, MTR carried 2.5 million passengers daily and the capacity of passengers was expected to grow. To enhance customer service, MTRC decided to increase train frequency during its peak rush hours.

Unfortunately, due to complexity in manipulating train movements within the depot interlocking area, it appears practically impossible for a human planner to produce an optimized timetable ? software optimization techniques were needed. MTRC issued a tender to study the scheduling of its trains in 1995.

AI Technology

Using advanced AI scheduling technology, we designed a timetabling algorithm that captures constraints related to train movements and models the operations within the MTRC depot. This algorithm produced a train timetable for the tightly congested depot interlocking area.

This type of train movement schedule was never automated before due to the complexity of constraints involved. The algorithm automatically generates a train timetable to dispatch and reverse trains based upon desired train headway at different periods. It also ensures the resulting timetable is operationally feasible by generating a train movement schedule that supports the timetable.

Business Benefits

Since delivering our software for the Train Scheduling System to MTRC in 1995, train frequency was increased during rush hours to improve their customer service. This is inline with their commitment towards a continuous improvement and a pressure relief on roads in Hong Kong.

More Info:

• Whitepaper - Chun, H.W., "Train Timetable and Route Generation Using a Constraint-Based Approach," Proceedings of the Tenth International Conference on Industrial & Engineering Applications of Artificial Intelligence & Expert Systems, Atlanta, June, 1997.

Last modified: Thursday, 28-Apr-2005 16:39:12 HKT

Copyright © 1994-2009, Dr. Andy Chun, Hon Wai, All Rights Reserved.      Privacy Statement