AI 中文总结
该研究提出统一优化框架，通过列生成启发式方案联合设计多模式交通的线路、频率与按需出行段，在波士顿、芝加哥数据集上较基准方案显著提升乘客量。
AI 中文摘要
固定线路公共交通与按需出行服务的融合，为大规模网络设计带来了建模挑战与计算机遇。我们提出一种基于流的混合整数规划模型，在固定运营预算下联合优化交通线路规划与服务频率，同时通过按需服务明确覆盖首英里与末英里的连通性。为实现城市规模下的可求解性，我们开发了一种新型列生成启发式方案，配备定制化定价子问题。将该框架应用于波士顿与芝加哥的网络及需求数据，得到了运营可行的设计方案，可大幅提升服务覆盖的需求规模。与仅公共交通、仅按需服务的基准方案相比，波士顿的乘客量分别提升最高达20.99%与93.58%，芝加哥则分别提升最高达10.63%与149.84%；与多模式基准方案相比，波士顿与芝加哥的乘客量分别提升5.42%与5.80%。这些结果表明：（1）在统一优化框架内联合设计交通线路、频率与按需出行段，在同等预算约束下，比单模式或解耦方法能带来显著更高的乘客量；（2）所提出的模型与列生成定价方案，相对于测试的基准方案可实现可求解、运营可行且性能优异的解决方案。
英文摘要
The integration of fixed-route public transit and on-demand mobility services presents both a modeling challenge and a computational opportunity for large-scale network design. We propose a flow-based mixed-integer programming formulation that jointly optimizes transit line planning and service frequencies while explicitly capturing first- and last-mile connectivity via on-demand services, under a fixed operating budget. To achieve tractability at urban scale, we develop a novel column generation heuristic scheme with tailored pricing subproblems. Applied to networks and demand in Boston and Chicago, the framework yields operationally feasible designs that substantially increase demand served. Relative to transit-only and on-demand-only baselines, ridership increases by up to 20.99% and 93.58% in Boston, and by up to 10.63% and 149.84% in Chicago. Compared to a multi-modal benchmark, our approach improves ridership by 5.42% and 5.80% in Boston and Chicago, respectively. These results demonstrate that (i) joint co-design of transit routes, frequencies, and on-demand legs within a unified optimization framework yields substantially greater ridership than single-mode or decoupled approaches under equivalent budget constraints; and (ii) the proposed formulation and column generation pricing scheme admit tractable, operationally feasible, high-performing solutions relative to tested baselines.
