A Novel PSOEDE Algorithm for Vehicle Scheduling Problem in Public Transportation

Abstract

One of the problems in public transportation is the vehicle scheduling problem (VSP), which can reduce the bus company cost and meet the demand of passengers’ minimum waiting time. This paper proposes an ensemble differential algorithm based on particle swarm optimization (abbreviated as PSOEDE) to solve the VSP. In PSOEDE algorithm, the mutation process is designed by dividing the original process into two parts: the first part combines the PSO operator with the improved mutation strategy to enhance the global search ability, while the second part is to randomly select two mutation strategies (i.e. random learning and optimal learning) to improve the diversity of population. In addition, the random selection methods of the parameters and crossover strategies are proposed and applied in the total PSOEDE algorithm. The effectiveness and superiority of the proposed PSOEDE algorithm in dealing with the VSP are verified using the simulation experiments and six comparison algorithms.