Timetable Synchronization of Mass Rapid Transit System Using Multiobjective Evolutionary Approach

Users of mass rapid transit are often required to make transfers between different train lines to reach their destinations. Timetable synchronization minimizes delays during such transfers. This paper formulates a novel measure for timetable synchronization by means of a total passenger dissatisfaction index (TPDI); and the impact of such synchronization on the original unsynchronized timetable is accounted using a total deviation index (TD_V) that assigns penalties when deviations from the original timetable are incurred. Pareto fronts displaying the relationship between TPDI and TD_V are generated using the state-of-the-art nondominated sorting genetic algorithm 2 (NSGA 2). To further improve NSGA 2, three schemes-the use of a variant of the NSGA2 with differential evolution, a process we termed "seeding," and finally a hybrid combination with local search techniques like heuristic hill climbing, tabu search, and simulated annealing - are proposed. Simulation results demonstrate that the "seeded" NSGA2-DE combined with the hill climbing heuristic produce the best results for the application. Solutions from the Pareto fronts are chosen for implementation to describe the different operating regions. A discussion section details the advantages and drawbacks of the proposed schemes.