随机时变路网下基于乘客时空灵活性的需求响应公交调度优化研究

需求响应公交旨在提供柔性半集约的出行服务，是出行需求多样化和信息化发展的产物，而这种离散化的服务模式不仅成本较高，在随机时变路网环境下还面临运营可靠性的挑战。现有研究基于固定的时空位移出行需求探讨运营调度优化，忽略了人的可移动性和车次延误传播效应；在算法上主要采用启发式算法获得可行解，不能保证全局最优。.与传统研究仅追求车辆管控的方法不同，本项目提出乘客时空灵活性的新机制，借助精细化预约管理释放需求侧灵活性降低系统成本和缓解路网不确定性的影响。运用方式选择理论标定时空灵活性成本，利用马尔可夫链模型推导车次延误传播效应。针对车次任务和衔接方式的复杂性，建立通用的时空网络模型，定义车流和客流时空网络图，在分层处理中嵌入灵活性成本；根据问题特征进行模型变换和高效精确算法设计，并通过字典序排序方法减少时空调整范围，实现系统成本与乘客偏好的多目标权衡。研究成果有益于借助移动互联技术发展需求响应公交。

Demand responsive transit aims to provide semi-flexible travel service, which adapts for travel demand diversification and information technology development. However, such a discrete service pattern not only has large cost, but also encounters the challenge of operational reliability under stochastic time-varying road environment. Existing research investigated schedule optimization based on the constant time-space movement travel demand, while neglecting the passengers’ mobility and trip delay propagation effect. In the aspect of algorithms, most of the existing works used heuristic algorithm to explore feasible solutions, which cannot guarantee the global optimum.. Distinctly from traditional research that focuses on the management and control of vehicles, in this research, we propose a new mechanism of passengers’ temporal and spatial flexibility, with an aim to reduce system costs and mitigate the impact of uncertainty by releasing demand-side flexibility via refined booking management. The cost of temporal and spatial flexibility is tuned by choice modeling theory, and the trip delay propagation effect is derived using Markov chain model. In view of the complexity of trip tasks and connection, the generalized time-space network model is developed. The vehicle-flow time-space network and passenger-flow time-space network are defined, and the cost of temporal and spatial flexibility is incorporated in the hierarchical process. According to the problem characteristics, model transformation and efficient exact algorithm design are carried out. The range of time-space adjustment is further reduced by lexicographic method, in an aim to trade off the system cost and passengers’ preference. This research will help develop demand responsive transit through mobile Internet technology.