求解多目标区域公交车辆调度问题的遗传算法

Genetic Algorithm for Multi-Objective Regional Bus Scheduling Problem

将区域公交车辆调度(RBSP)视为"部分班次被某车辆完成"的集合划分问题,考虑站场容量、补充燃料和返回站场最迟时间限制等约束因素,建立一类追求车辆利用效率和公交车辆调度方案的可靠度极大化的多目标混合整数规划模型.用约束法将该模型转为单目标规划问题,利用遗传算法求解该问题的非劣解.研究表明:公交调度方案的可靠性越高,车辆利用率越低.

Regional bus scheduling, treated as a ＂part of trips completed by a vehicle＂ set-partitioning problem, ould be formulated as a multi-objective mixed integer programming model, where some side constraints such as depots capacities, fuelling and the latest time of going back to depot were considered. The previous objective was to maximize vehicle utilization for all buses, and the secondary objective was to maximize reliability of scheduling plans. Before the model＇s non-inferior solutions were obtained by genetic algorithm, it was converted into a single objective programming problem with constraint method. Finally, an example shows that the high reliability of schemes become, the less vehicle utilization will get.