A Genetic Algorithm for Bus Schedule Synchronization

The bus transfers synchronization problem is to find an optimal bus schedule for the entire transit system that minimizes riders' transfer time. This is a complicated problem because of the large set of binary and discrete values involved. The combinatorial nature of the problem poses a computational burden and makes it difficult to solve using classical methods. This paper presents a Genetic Algorithm (GA) approach to synchronizing bus schedules to minimize transfer times. The algorithm takes existing scheduled timetables and ridership data at transfer locations as input and addresses randomness in bus arrivals. The algorithm attempts to find an optimum solution for the bus schedule synchronization problem by shifting existing timetables. Using scheduling data from Broward County Transit, total passenger transfer times were calculated for the existing and the proposed systems. The algorithm was able to reduce system-wide transfer time by 13.4%. Results from this case study and recommendations for future research are presented.