刘心雨
入学时间：2017级
答辩时间：2021年
论文题目：面向服务的路面公交线路优化方法研究
中文摘要
摘要
在我国，路面公交将在长期范围内承担城市交通运输的主体作用。因功能丰富，加之路面运行干扰较多，其服务具有复杂性。近年来，路面公交面临前所未有的严峻竞争。需求侧，人们对出行品质的要求提高，追求多元舒适的出行体验。供给侧，新能源汽车和自动驾驶技术为小汽车赋能，克服了无法利用在途时间、依赖于驾驶技术、高能耗等缺点；网约车、共享单车等新兴交通方式逐渐占据市场，发展成为一种稳定可用的“他服务”交通方式。区别于以往对公交系统整体优化，本研究创新性地讨论路面公交服务本质、要素、系统及其有机关系，围绕公交线路新建调整、运营服务、评价诊断的全生命周期，建立了一整套融入政府、公交企业、出行者及环境交互的公交线路优化方法。
公共交通系统是由人、车、线、站、环境等共同构成的有机整体，本研究以路面公交为典型对象，提出了一般意义上的服务理论基础，包括：（1）公共交通服务本质及面向服务的公共交通系统概念；（2）公交服务内涵，即以政府和公交企业为服务提供者、以乘客为服务接受者、三方交互形成的服务接触过程；（3）公交服务实体、要素、服务质量属性及其有机关系；（4）公交服务运营模式、价值链及系统架构。由此提出了以服务指导公交线路优化的思路：考虑政府、公交企业和出行者之间的交互过程，在线路新建调整阶段，使政府以合理补贴激励公交企业提供服务、企业的运营资源投入与使用公交的需求平衡；在线路评价改善阶段，评价运营成本投入转化为经济性和质量性服务产出的效率，诊断低效率线路的调整优化方向及政府、企业责任，以便投入下一次运营。随后阐述了基于多源数据估计客流需求和提取运营指标的数据分析方法，为线路优化提供基础输入。
面向政府难以调动企业积极性、公交线路供需不均问题，针对单线路公交建立了以政府为引导者、公交企业为跟随者的新建调整优化方法。考虑我国公交服务的主要运营模式及政府和公交企业之间的决策关系，基于斯塔克尔伯格博弈（Stackelberg Game）构建双层优化模型，揭示政府、公交企业和出行者三方的交互过程。上层政府作为公交线路优化的引导者，以社会总成本最小化为目标决定对公交线路乘客的补贴额度；下层公交企业作为跟随者，在给定补贴额度情况下，以线路盈利最大化为目标确定公交线路的发车频率和配车数。优化时，政府通过构建公交企业的反应函数预测企业在不同补贴额度下的最佳决策，从而确定使社会总成本最优的补贴额度。出行者基于公交线路的发车频率及路网中的其他可用交通方式进行方式和路径选择，出行者的选择将会影响上层政府和下层企业的决策目标。通过分析公交企业对政府补贴的反应规律，设计了多区间二分法对模型进行求解，并以上海市公交85路为例验证了模型和算法的效果。
拓展单线路公交中政府、公交企业和出行者之间的交互过程，以线路组或小范围公交线网为研究对象，建立了以政府为引导者、公交企业为跟随者的公交线路组新建调整优化方法。上层政府以社会总成本最优为目标决定分配给公交乘客的补贴额度，下层公交企业以线路组总盈利为目标决定线路的空间布局、发车频率及配车数。考虑多条直达或有1-2次换乘的可用公交路径以及其他可用交通方式，建模出行者的方式和路径选择，作为上下层优化的基础。结合上、下层问题特性，提出了混合遗传模拟退火算法求解，上层使用模拟退火算法对最优补贴额度进行搜索，下层基于遗传算法生成对应于特定补贴值的线网布局、发车频率和配车数最优解。以已有研究的道路网络和需求矩阵为例，对比探索了不同场景下算法的优化效果。
面向公交服务评价与改善脱节问题，建立了公交线路服务绩效评价与诊断分析方法。应用能够自动分配指标权重、并对多个决策单元统一评价的数据包络分析（Data Envelopment Analysis）模型及其拓展，基于多源公交数据分析，建立了融入政府、公交企业和环境因素的服务绩效评价模块，帮助公交企业定期评价公交线路，并设计了针对低效率问题线路的诊断机制，从而为基于评价结果的公交线路优化提供方向。在服务绩效评价部分，将绩效评价分为技术效率和服务效益，前者评估公交线路运营投入转化为服务产出的效率，后者考虑出行者的参与过程，评估服务产出转化为服务消费的效率，在评价中同时选取经济性指标和服务质量指标，涵盖政府、公交企业和环境因素，以便划分责任。基于线路的技术效率和服务效益评价结果，考虑长期改善和短期调整两种情况，应用相应诊断模型，生成问题线路的优化策略和改善方向，辅助公交企业实施改善措施。研究以济南市中心城区的23条常规公交线路为例，说明了公交线路的评价诊断过程。
本研究所提出的服务理论基础，揭示了路面公交的服务本质及其内涵、要素、属性与架构，该理论也可拓展至其他交通方式，为由政府、交通企业和出行者及其交互过程构成的复杂交通系统服务优化提供参考；围绕单线路公交和小范围线网所构建的新建调整优化方法，充分考虑了政府、公交企业和出行者三个主体的决策目标、内容及其交互，通过实证分析可知，该方法能够帮助政府有效引导公交企业提供服务，促进公交线路服务的供需平衡，降低社会总成本；公交线路评价诊断分析方法，能够以服务绩效评价指导公交线路优化调整，打通评价和改善链条，实现对公交线路的闭环、可持续优化管理。
关键词：路面公交，公交线路优化，服务绩效，面向服务
英文摘要
ABSTRACT
In China, bus transit will remain a primary urban transportation mode in the long term. However, in recent years, bus transit is facing severe challenges. People have increased their standards for the quality of traveling due to economic growth, shifting to more diversified and comfortable transportation modes. On the other hand, the development of electric cars and autonomous vehicles helped to overcome some disadvantages of automobiles, including failure to utilize in-vehicle time, requiring driving skills, high energy consumption, etc. Travel experience with automobiles has been upgraded. Another challenge comes from shared mobility and shared bikes which have steadily occupied the market for the past decade. Therefore, to attract more passengers and increase the market share of buses, the provision of bus transit needs to be revolutionized. This study innovatively discusses the service nature, elements, system structure along with relations between system components of bus service, seeking to establish a set of bus route optimization methods for the life cycle of bus routes, including route establishment, route adjustment, service operation, service evaluation, and service diagnosis, detailed as follows:
Public transport system consists of people, vehicles, routes, stops, and environment. Using buses as an example, this study proposes the fundamentals of transit service, including: (1) the service nature of public transport and the concept of service-oriented public transport system; (2) the connotation of bus service, i.e., service encounters between providers (governments and transit agencies) and recipients (passengers); (3) bus service entities, components, attributes along with their relations; (4) bus service operation models, co-creation of service values, and system structure. The ideas of service-oriented bus route optimization are proposed: (1) establish optimization methods for a single bus route and a set of bus routes with considering the interactions among governments, transit agencies, and travelers at the route establishment and adjustment phase; (2) build a diagnosis-oriented and evaluation-driven framework for bus route improvement at the route evaluation and improvement phase. Following these ideas, methods are developed to estimate transit demand and extract operation parameters from multiple types of transit data to provide inputs for further optimization.
To solve the problem that governments fail to encourage transit agencies in providing better services and balance bus demand and supply, a Leader-Follower Stackelberg-game-based single bus route optimization method is established for setting up a new route or adjusting an existing one. Considering the operation models of bus service in China, especially the relationship between governments and transit agencies, the proposed model features a bi-level structure with the upper level reflecting the perspective of government agencies in subsidy allocation and the lower level representing the decisions of service providers in dispatching frequency and bus fleet size design. The bi-level model is framed as a Stackelberg game where government agencies take the role of “leader” and service providers take the role of “follower” with social costs and profits set as payoffs, respectively. The leader’s decisions are optimized with the prediction of the follower’s reaction. Decisions of fleet size and frequency by the transit provider would change the cost of the bus route and influence mode and route choices of travelers, resulting in changes of objectives for transit providers and the government accordingly. A partition-based bisection algorithm is developed to solve the proposed model. Results from a case study of Bus Route 85 in Shanghai validate the effectiveness and performance of the proposed model and algorithm.
A Leader-Follower Stackelberg-game-based bus route network optimization method is proposed to help establish or adjust a set of bus routes or a small network. Extended on the single bus route optimization framework, the upper level reflects the perspective of government agencies in subsidy allocation, and the lower level represents the decisions of service providers in network design, frequency setting, and bus fleet size. Travelers’ available bus route options include direct transit trips, trips with one transfer, and trips with two transfers. An integrated Simulated Annealing-Genetic Algorithm (SA-GA) is proposed to solve the problem. The upper-level objective is optimized by Simulated Annealing algorithm, and the lower-level optimal solution is searched by modifying Genetic algorithms. The proposed method is validated through comparisons on a numerical case from previous research. Results are tested under different scenarios.
A diagnostic-oriented and evaluation-driven framework is proposed to bridge the gap between bus route evaluation and improvement. Based on the extensions of Data Envelopment Analysis (DEA) models, the roles of government agencies, transit operators, and passengers are incorporated in the route performance evaluation module to help transit service providers with their day-to-day operation. A diagnosis scheme is further developed to target inefficient bus routes and provide improvement directions. In the evaluation module, input and output variables are extracted from multiple data sources to evaluate technical efficiency and service effectiveness of bus routes, which focus on the efficiency of transferring operation inputs to service outputs and service outputs to service consumption, respectively. Both quantitative and qualitative measures are used for evaluation, which covers the roles of governments, transit service providers, and environmental factors. Governments and transit service providers could then split responsibilities from each other. After evaluation, the inefficient bus routes are analyzed through an improvement diagnosing process featured with route categorization and improvement direction identification. Recommendations are given based on the evaluation results and long-term or short-term improvement purposes, to assist transit agencies in developing practical solutions. The framework is applied and validated in a case study for service performance improvement of 23 bus routes in Jinan.
The transit service fundamentals illustrate the nature, connotation, attributes, and system structure of bus service, which can also be extended to other transportation modes, providing recommendations for complex transportation systems with the nature of service encounters between governments, agencies, and passengers. The bilevel optimization methods for a single bus route and a small bus network give profound considerations on the decisions of governments, transit service providers, and travelers along with their goals and interactions during bus route establishment and adjustment. Through case studies, the proposed methods could optimize governments’ subsidy decisions to encourage transit agencies in providing service and balance bus service demand and supply, minimizing social cost efficiently. The diagnostic-oriented and evaluation-driven framework can bridge the gap between transit performance evaluation and improvement prioritization at the route level, which provides a closed loop of bus route service for transit agencies to continuously improve bus service operation.
Key Words:bus transit, bus route optimization, service performance, service oriented
