Optimal operation for rail transit systems under advanced information

Rail transit is normally operated on a fixed train schedule (timetable), designed based on data from typical days. In practice, however, unexpected fluctuations in passenger flow and/or in facilities may occur, making the original schedule unrealizable or non-optimal. This calls for a real-time Decision Support System (DSS) which can assist transit operators to effectively adjust the train schedule on the real-time basis when the operation environment changes markedly. Such a system can be made possible by the latest developments in intelligent transportation technologies. As the theoretical part of an operational DSS, this paper presents an optimization model, based on information available from the advanced surveillance technologies (e.g. the current situation of facilities, the short-term prediction of passenger flow, etc.), to optimize the real-time train schedule for a specific time horizon. An approximation algorithm for this model is proposed and some computational results are reported.