This paper studies a vehicle relocation problem for one-way carsharing systems in which vehicles are relocated by temporary workers during the night. The goal is to develop a method for constructing employment plans and work schedules simultaneously to minimize the total relocation cost. An integer programming model is proposed, which incorporates employment planning, return restrictions and ride sharing of temporary workers. An iterative optimization approach is developed to address large-scale rebalancing problems. The unbalanced stations are clustered based on their relocation demands. Considering a new employment limit in each rebalancing model, the optimization procedure iteratively solves a set of cluster relocation problems and an inter-cluster relocation problem until a convergent solution is obtained. Numerical experiments based on data from a carsharing company in Chengdu reveal two interesting findings: (1) Employing temporary workers from understocked stations costs less than that from overstocked stations; (2) Using a time-based salary contributes to 49.1% reduction in the working hours and 33.8% in the relocation costs.

本文研究了单向汽车共享系统的车辆搬迁问题，其中临时工在夜间搬迁车辆。目标是开发一种同时构建就业计划和工作时间表的方法，以最大限度地减少总搬迁成本。提出了一个整数规划模型，该模型结合了就业规划、返回限制和临时工的拼车。开发了一种迭代优化方法来解决大规模的再平衡问题。不平衡站根据其搬迁需求进行聚类。考虑到每个再平衡模型中的新就业限制，优化过程迭代地解决一组集群迁移问题和集群间迁移问题，直到获得收敛解。基于成都一家汽车共享公司数据的数值实验揭示了两个有趣的发现：（1）从库存不足的站点雇用临时工的成本低于从积压站点雇用的成本；(2) 采用计时工资制，工时减少49.1%，搬迁成本减少33.8%。