不确定需求下基于分布式鲁棒机会约束的车辆调度问题研究

随着新能源汽车共享模式的迅速发展,允许异地还车导致车辆不均衡问题日益突出.本文考虑不确定的车辆需求,基于需求量的均值和方差等部分信息,最小化最坏情况下系统可用车辆和空闲停车位的机会约束,建立分布式鲁棒优化机会约束车辆调度模型,以总成本最小化为目标,得出模型的数学性质和易求解的等价形式,确定停车桩之间的车辆调度数量.最后,以北京市15个停车桩的车辆调度为例,确定最优的车辆调度方案.结果表明,随着服务水平的增加,调度成本和车辆调度数量增加明显,当服务水平增加到一定程度后,应增加系统的车辆数和停车位以满足服务水平的要求.决策者可根据自己的偏好及系统的服务要求,选择恰当的服务水平参数组合,以获得最优的车辆调度方案.

Distributionally robust chance-constrained vehicle scheduling with uncertain demand

With the rapid development of the new energy vehicle sharing system, the imbalance problem became increasingly prominent due to car returning in different sites. We consider uncertain demand. Based on mean and variance of demand distribution, we propose distributionally robust chance constraint vehicle scheduling model to minimize the chance constraints of the worst available vehicle and free parking spaces, then this model can be formulated as tractable counterparts. Finally, we apply this approach to the case of service with real operations data in Beijing. Numerical results show that, with the increase of service level, the scheduling cost and the number of vehicle increase obviously. When the service level improves to a certain extent, the number of vehicles and parking spaces should increase to meet the requirements of the service level. According to their risk preferences, decision-makers can select the appropriate level of service parameters to get the optimal vehicle scheduling.