基于多源公交数据和车时成本优化的公交运营时段划分方法

现有公交运营时段划分方法均基于单一参数（乘客需求或行程时间）的相似性，忽略了乘客需求与行程时间对时段划分方案的协同作用。针对传统方法的缺陷，利用多源公交数据，提出一种新的基于最小化车队运营时间成本的时段划分方法。首先，利用逆差函数模型计算时间窗内完成班次任务所需要的最小车队规模，进而利用滑动时间窗模型，计算全天各时间窗所需的理论最小车队规模。然后，以时段划分方案的时间分割点为决策变量，以最小化全天累计车队运营时间成本为目标建立优化模型，采用遗传算法对运营时段划分方案进行寻优。最后，以广州市87路公交线路实际数据为例进行验证，并对模型参数进行敏感性分析。研究结果表明：与传统方法相比，所提方法能更好地实现运力与运量相匹配，并有效降低车时成本；与以往基于断面客流和基于行程时间的划分方案相比，所提方案的车时成本分别降低25 veh·h和45.33 veh·h。

Time-of-day Interval Partition Method for Bus Schedule Based on Multi-source Data and Fleet-time Cost Optimization

Current methods for time-of-day interval partition problem are mainly based on the similarity of a single parameter (passenger demand or travel time) while neglecting the combined effect of these parameters on the resulting scheme. To overcome this defect, in this paper a new method for the time-of-day interval partition is proposed based on fleet-time cost optimization using multi-source data. Firstly, the deficit function model was used to calculate the minimum fleet size required to fulfill the trip tasks of a time window. Subsequently, with the moving time window, the theoretical minimum fleet size in each time window during the operation period was calculated. Taking the time interval partition plan as the decision variable, the optimization model was developed with the objective of minimizing the cumulative fleet-time cost of the fleet throughout the day. The genetic algorithm was used to search the optimal time interval partition plan. Finally, the data from Route-87 bus in Guangzhou, China was used as an example for validation, and sensitivity analysis was conducted for the model parameters. The results show that compared to the existing approaches, the proposed method can better realize the matching between fleet capacity and passenger demand, and effectively reduce the fleet-time cost. Compared to traditional schemes based on either passenger flow or travel time, with this method the fleet-time cost is reduced by 25 veh·h and 45.33 veh·h, respectively.