需求响应式公交系统车辆调度问题的交互式优化方法研究

In recent years, with the widespread use of smartphone applications, a new and innovative mode of cyber-enabled, demand-responsive transit (DRT) systems has been springing up across the world. The vehicle scheduling methods used in conventional transit systems cannot be directly applied to solve the vehicle scheduling problem of the new DRT systems. This research project focuses on the vehicle scheduling problem of the new DRT systems, and it is three-fold: (i) developing methods for calculating the minimum fleet size required for the new DRT systems, (ii) developing static vehicle scheduling methods for the new DRT systems, and (iii) developing dynamic vehicle scheduling methods for the new DRT systems. Based on the theories of network flow models, deficit function models and vehicle-shareability network models, a new mathematical programming-based advanced deficit function model will be developed for both static and dynamic vehicle scheduling of the new DRT systems. A two-stage human-machine interactive optimization method will be developed for solving the mathematical programming-based advanced deficit function model. Numerical experiments and case studies will be conducted to test the effectiveness of the model and optimization methods developed. It is anticipated that by utilizing the latest research results of human-machine interactive optimization method, the wisdom, experience and preference of DRT schedulers and users can be incorporated into the computerized optimization process via graphical user interfaces so as to develop a completely new, more flexible, more efficient, and more practical methodology for the vehicle scheduling of the new DRT systems.

近年来，随着智能手机App的广泛应用，一种新型的、网约式的、按需设计的需求响应式公交服务系统在全世界范围内迅速兴起。传统公交车辆调度方法不足以支撑此类新型公交系统的车辆调度。本项目围绕需求响应式公交车辆调度这一新问题，主要研究三个问题：(i)需求响应式公交系统最小车队规模问题的计算方法；(ii)需求响应式公交车辆静态调度方法；(iii)需求响应式公交车辆动态调度方法。本项目将综合运用网络流模型、逆差函数模型与车辆共享网络模型的理论知识，构建基于数学规划模型的高等逆差函数模型；设计两阶段人机交互式优化算法；并通过数值实验与案例分析对模型与算法的有效性进行验证。本项目希望借鉴人机交互式优化方法的最新研究成果，将调度员与乘客的智慧、经验与偏好等要素通过图形用户界面接口嵌入到计算机对问题进行优化求解的过程中，从而构建一套全新的、更加灵活高效的、更加符合现实所需的需求响应式公交车辆调度的新方法体系。

近年来，随着智能手机App的广泛应用，一种新型的、网约式的、按需设计的需求响应式公交服务系统在全世界范围内迅速兴起。传统公交车辆调度方法不足以支撑此类新型公交系统的车辆调度。本项目围绕需求响应式公交车辆调度这一新问题，主要研究了需求响应式公交系统最小车队规模问题的计算方法、需求响应式公交车辆静态和动态调度优化模型与求解方法。本项目综合运用网络流模型、逆差函数模型与车辆共享网络模型的理论知识，构建了基于数学规划模型与逆差函数模型相融合的人机协同决策模型，设计了两阶段人机交互式优化方法，并通过数值实验与案例分析验证了模型与算法的有效性。项目的研究成果为需求响应式公交车辆调度提供了新的理论与方法支撑。截至目前，共发表了11篇高质量学术期刊论文，其中2篇论文发表于交通科学与技术领域的顶尖期刊《Transportation Research Part C》、1篇论文发表于智能交通系统领域的顶尖期刊《IEEE Transactions on Intelligent Transportation Systems》，发表书籍章节1章、学术会议论文4篇。项目负责人在国内外学术会议做口头报告2次，牵头组织智慧公共交通领域的学术期刊专刊2次、参与组织学术会议专栏3次。申请发明专利3项。培养在读硕士研究生5人、本科毕业论文（设计）6人。项目研究成果服务于我国智慧公共交通领域的建设发展需求。