A Microscopic Modeling Approach for Real-Time Train Retiming Under Disturbances for a Cbtc Suburban Railway Line

To cope with the increasing transport demand in big cities, automated systems such as the Communication Based Train Control (CBTC) system are now deployed on dense and open suburban lines. The increase in traffic density allowed by these systems makes the railway lines more sensitive in case of disturbances. To manage a line under disturbances, this paper addresses the train timetable real-time retiming problem designed to determine updated dwell times and departure times for CBTC trains, while minimizing the drift from the planned timetable in terms of both punctuality and regularity of headway. To be consistent with the impact of the disturbances on the planned timetable and the high level of detail at which decisions must be taken, a microscopic model of the railway infrastructure is proposed in this paper. Reassignment and resequencing decisions are represented (resp. instantiated) on a disjunctive (resp. conjunctive) graph. Industrial constraints including signaling, headway, and sharing the common resources, are considered. Based on empirical data, a piecewise linear regression model is proposed to determine the inter-station running times to efficiently monitor and measure the impact of disturbances on the performance of CBTC trains on the studied line. The train timetable retiming problem results in a mixed-integer linear program with quadratic terms in the multi-objective function. The effectiveness of the proposed optimization model on industrial data are assessed based on a simulation-optimization approach.