Train routing and timetabling via a genetic algorithm

Abstract

Train timetabling is a complex and time-consuming problem, which must be solved accurately to gain maximum performance on using the resources of a railway company (railways, trains, staff, etc.). An appropriate time bill (a timetable) must have a low latency, suitable train routes and a minimum time for traveling passenger trains and freight trains. In this paper, a genetic algorithm for routing and scheduling trains is developed to achieve efficient and robust train routes and a timetable. The algorithm may change the start times of the trains with the aim to find better times for the trains to start their moving and to have less number and a smaller total time of possible train delays. The algorithm changes some parts of the train paths to the destination to have a less mutual exclusion to use railways if it is allowable for a train (e.g., freight trains and express passenger trains are not allowed to stop at some middle stations of their routes). Simulations on a computer show that the developed genetic algorithm can reduce the tardiness of the trains to go from a source station to a destination.