作为城市公共交通运输系统中的创新服务模式，需求响应式公交服务具有个性化、便捷化、共享化等特点。由于尚处于探索发展的初期阶段，需求响应式公交的规划、部署、应用与发展亟需优化战略和理论方法支撑。本论文以“建设人民满意交通”为发展目标，针对需求响应式公交服务设计和运营优化过程中的难点，从“战略-战术-运营”层面探索、解析和制定需求响应式公交的全过程发展路径。基于深圳市出行实例，验证了本论文所提出模型与算法的有效性，为我国城市公共交通运输系统高质量发展提供支撑。主要研究内容包括：（1）基于需求响应式公交在实际应用中需要与传统公共交通密切配合、协同发展的战略定位，选取地铁作为代表交通方式阐述了需求出行模式的分类逻辑，即地铁竞争型与地铁合作型出行。结合GPS轨迹大数据，对不同出行模式需求的时空分布特性进行精细至路段、具体至时段的精准画像，采用基于筛选样本的二元Logit模型进行影响因素分析。案例结果显示，地铁竞争型出行需求存在显著的时空差异，且路网结构、交通状态和建成环境因素对其均具有显著影响。本论文提出了分析视角、时空尺度和影响因素更为全面的出行模式研究框架，解析了不同模式需求的出行行为机理，为因地制宜地提供需求响应式公交服务场景奠定了基础。（2）面向地铁合作型出行需求，以乘客提交的出行信息和地铁时刻表为输入，构建需求响应式公交与地铁的一体化预约出行服务模型，为乘客提供全过程出行方案。综合运营商和乘客利益，提出包含运营成本、时间成本和失效成本的综合成本目标函数，构建车辆调度与服务优化模型，并采用模型线性化处理方法和拓展的自适应大领域搜索算法进行求解。结果显示，该模型能够综合多主体的利益诉求，较传统的单一主体利益最优导向的需求响应式服务效率更高。本论文从乘客全出行链一体化服务的视角出发，为多模式组合出行服务提供一体化优化决策支撑。（3）面向地铁竞争型出行需求，引入换乘机制，建立了面向实时动态需求的需求响应式公交一体化动态服务模型。该模型以最大化运营收益为目标，综合考虑服务收益、运营成本、失效成本以及车辆折旧成本等要素对运营收益的影响，并考虑了协同换乘等约束条件，保证了乘客的服务质量。同时，将静态模型与滚动时域搜索框架集成以实现需求的实时响应，案例结果证明该系统能够提高服务效率和运营效益。本论文结合换乘机制和实时响应特质，为换乘便捷协同、车辆的高效调度提供了一体化决策方法。
As an innovative travel mode within urban public transportation systems, Demand-Responsive Transport (DRT) services are characterized by their personalization, convenience, and shareability. Being in the initial stages of exploration and development, the planning, deployment, application, and advancement of DRT urgently require optimized strategies and theoretical methods for support. Aiming at the development goal of ‘Buliding People-Oriented Transporation’, this study explores, analyzes, and formulates a comprehensive development pathway for DRT services from strategic, tactical, and operational perspectives, addressing the challenges in the design and operational optimization process. Based on travel examples from Shenzhen, the effectiveness of the models and algorithms proposed in this study is verified, providing support for the high-quality development of China‘s urban transportation systems. The main research contents include:(1) Based on the strategic positioning that demand-responsive transit must closely coordinate and develop with traditional public transportation, the subway is selected as the representative mode of transport to elucidate the classification logic of demand travel patterns, namely subway-competitive and subway-collaborative demand. Utilizing big data from GPS tracking, this research finely maps the spatiotemporal distribution characteristics of different travel demand patterns down to specific road segments and time slots. A Selected-sample-based Binomial Logit model is used for the analysis of influencing factors. Case study results reveal significant spatiotemporal differences in subway-competitive travel demand, with network structure, traffic conditions, and the built environment having a significant impact. This study proposes a research framework for travel pattern analysis that encompasses broader perspectives, spatiotemporal scales, and influencing factors, laying the foundation for differentiated service scenarios in demand-responsive transit tailored to various travel needs.(2) For subway-collaborative travel demand, considering the integrated service of demand-responsive transit with the subway, a full-process travel solution is provided based on passengers‘ travel information and subway schedules. Starting from the interests of operators and passengers, a comprehensive cost objective function based on operational cost, time cost, and failure cost is proposed. A vehicle scheduling and service optimization model from a passenger‘s whole journey perspective is constructed, and solutions are sought through linearization of the model and an extended adaptive large neighborhood search algorithm. Results show that this model can integrate the interests of multiple stakeholders more efficiently than traditional single-perspective demand-responsive services, offering an integrated optimization decision-making method for multimodal travel services.(3) For subway-competitive travel demand, the coordinated transfer is introduced, establishing an integrated service model for demand-responsive transit for real-time dynamic demands. The model aims to maximize operational revenue by comprehensively considering the impacts of service revenue, operational costs, failure costs, and vehicle depreciation on operational revenue, and also takes into account constraints such as coordinated transfers, ensuring the quality of service for passengers and the interests of operators. Furthermore, the integration of the static model with a rolling horizon search framework enables real-time demand response. Case results demonstrate the system‘s potential to improve service efficiency and operational benefits, providing an integrated decision-making method for real-time service response, convenient transfer coordination, and efficient vehicle dispatch.
