Abstract
In railways, fixed timetables are provided day by day at present, meaning that those timetables do not fit passengers' demand of each occasion. In this paper, we propose an idea to dynamically change train timetables reflecting passengers' demands so that passengers can be provided with better transportation services. To realize this idea, we introduce a train timetabling algorithm which organizes train timetables directly from passengers' demands. Unlike conventional works, the algorithm takes all passengers' disutilities into account by estimating each passenger's behavior towards a train timetable, and produces a timetable in which passengers' disutilities are minimized and efficiency of resource utilization is maximized. Hence, this algorithm takes passengers' demands as its input and outputs a near optimal timetable from the viewpoints of the criteria described above. We have conducted several experiments using the data which were made from actual railway lines and shown that the algorithm is capable of producing timetables which properly fit the passengers' demands.
