王进

入学时间：2008级

答辩时间：2012年

论文题目：信号控制交叉口群时空协调设计基础问题研究

张毅

入学时间：2008级

答辩时间：2011年

论文题目：基于多维影响因素的通勤出行决策行为

中文摘要

摘要

随着城镇化、机动化水平的提高，我国城市居民的日常活动范围不断扩大、个体机动化出行趋势日益加剧，对有限的城市道路资源带来巨大压力，引发城市交通的诸多问题。为此，我国政府提出了“优先发展城市公共交通”的战略，意在调控和引导出行行为、平衡交通需求和供给。但在实际操作中，以公共交通发展为导向的交通需求管理在我国尚处于起步阶段，战略层面的“公交优先”并没有转化成战术层面的交通需求管理措施，公交方式在出行方式结构中的比例不升反降情况有之。造成这一局面的原因非常复杂，其中重要原因之一是我国出行决策行为的理论研究对实践应用尚缺乏有效支撑。

关于出行决策特征，家庭或个体的出行决策行为包含若干连续的链状出行行为，从居住\工作地点选择和交通工具拥有等长期决策行为，到出行生成和出行方式选择等短期决策行为，符合行为决策理论中连续决策的特征。对出行决策行为展开研究，从决策过程角度更真实地模拟一系列出行行为，能够突破以往仅仅关注单一出行行为的局限性。

关于出行影响因素，出行决策行为受到家庭、个人、交通系统、土地利用和出行意愿等多个维度因素的影响。建成环境（Built Environment）整合了土地利用、道路设施、交通服务等方面的物质环境，以及该物质环境中的人类活动模式，在研究中引入建成环境特征的因素能够为解释出行决策行为提供新的视角。同时，解决通勤时段的交通问题对改善现有的城市交通具有关键影响。

因此，为了解析通勤出行决策行为，围绕上述问题及其有机关系，基于家庭、个人和建成环境特征维度的因素与通勤出行决策行为的显著相关性，本文展开了“多维影响因素——通勤出行决策行为”的研究。研究的主要内容和结论包括：

（1）研究的核心是解析多维影响因素与通勤出行决策行为之间的内在关系。为此，本研究：①引入了建成环境特征的影响因素，与常用的家庭、个人特征影响因素共同构成多维影响因素；②选择了三种具有一定连续性和相关性的通勤出行决策行为作为研究对象：家庭交通工具拥有行为、家庭分方式通勤出行生成行为和个体通勤出行方式选择行为；③确定科学合理的“多维影响因素——通勤出行决策行为”研究方法框架体系，包括“收集数据®确定变量®建立模型®标定模型®解析结果”等五个环节；④并以沿海中等发达城市——中山市为案例对研究方法进行实证应用；⑤最后通过实证案例研究的结论对不同层面交通需求管理提出了政策、措施建议。本文采用多维影响因素突破了研究视角，选取三种通勤出行决策行为拓展了研究思路，构建方法框架体系丰富了研究手段，采集一手案例数据应用了研究方法，通过实证案例研究获得了实践反馈，上述创新点提升了研究的理论意义和实用价值。

（2）在确定变量环节，明确了建成环境特征指标体系，采用了基于GIS数据的多元统计分析方法将建成环境影响因素定量划分为特征指标、特征因子和邻里类型三个子维度，作为模型的自变量输入。首先通过GIS软件对建成环境原始数据进行加工获得特征指标，其次通过因子分析法将特征指标转化为特征因子，最后通过聚类分析法以特征因子为依据将研究范围内的邻里转化为若干具有代表性的邻里类型。根据该方法对中山市建成环境特征进行分析，确定了十种特征指标、五种特征因子和六类邻里类型。确定科学的建成环境特征定量研究方法，以我国城市为背景进行方法的应用，具有一定的创新。

（3）在模型建立环节，介绍了多元线性回归模型和非集计模型的建模方法、适用条件和选择方法。通过对因变量数据分布特征的检验，确定采用泊松回归模型模拟家庭交通工具拥有行为、采用负二项回归模型模拟家庭分方式通勤出行生成行为。通过对被选择方案“独立不相关”特性的检验，确定采用多项非集计模型模拟个体通勤出行方式选择行为。

（4）在模型标定环节，探讨了用于标定的统计分析软件和标定结果的统计特征分析方法。采用Stata 10.0软件，模型的标定结果清晰地表明了多维影响因素与通勤出行决策行为之间的定量关系，并证明了研究假设——家庭、个人特征维度的因素是通勤出行决策行为的主要影响因素，作用机理和力度比较稳定；建成环境特征维度的因素是次要影响因素，能够增加模型对通勤出行决策行为的解释能力。同时，对标定结果进行统计特征分析发现：研究对模型和变量的选取均科学合理。

（5）在结果解析环节，以弹性系数分析为依据，从定量和定性相结合的角度研究了多维影响因素的作用机理。家庭特征维度中，收入特征是最主要的影响因素，直接所用于家庭交通工具拥有行为，并间接作用于家庭分方式通勤出行生成行为和个体通勤出行方式选择行为。个人特征维度中，收入和机动化特征是主要的影响因素，直接所用于个体通勤出行方式选择行为。基于家庭和个人特征维度因素的发展趋势，中山市居民的通勤出行决策行为将呈现“个体机动化为主导”的特征，表现为：个体机动化交通工具逐步取代非机动化交通工具，部分非机动化或公交通勤出行转化为个体机动化通勤出行，出行者对个体机动化方式的选择概率压过非机动化或公交方式。

（6）建成环境特征维度中，在特征指标和因子层面，高密度和混合度的土地开发、便捷的目的地可达性等特征能够促进通勤出行决策中对非机动化方式和公交方式的选择，抑制对个体机动化方式的选择，具体表现为：更高的家庭非机动化交通工具拥有量、家庭非机动化或公交通勤出行生成量和个体非机动化或公交方式选择概率。发达的公交设施会促进通勤出行决策中对公交方式的选择，抑制对自行车、电动车或摩托车方式的选择；发达的道路设施会促进对个体机动化方式的选择，抑制对非机动化方式和公交方式的选择。在邻里类型层面，位于城市的两类邻里与步行方式和自行车方式的通勤出行决策行为均具有显著正相关性，位于城市和郊区的四类邻里与摩托车方式和小汽车方式的交通工具拥有行为和通勤出行生成行为具有显著正相关性，位于农村的两类邻里与公交方式的通勤出行决策行为具有显著正相关性。研究结论为从城市总体规划和城市综合交通规划层面制定有针对性的交通需求管理措施奠定了理论基础。

（7）在实践应用环节，在城市总体规划层面，提出保持土地开发的适宜强度、引导土地利用的混合开发和推动“工宿平衡”的格局构建等政策措施；在城市综合交通规划层面，提出重视

系统的规划设计、加快“微循环”系统的建设完善和贯彻多方式的综合与衔接等政策措施；在公共政策层次，提出形成城市小汽车拥有的动态配额机制、制定小汽车拥有和使用的费用政策、提供合理有效的出行信息等政策措施。

最后，在总结论文主要研究成果和创新点的基础上，指出了进一步的研究方向。

关键词：多维影响因素，通勤出行决策行为，建成环境，多元线性回归模型，非集计模型

英文摘要

ABSTRACT

Along with the rapid economic development, the levels of urbanization and motorization are increasing at a high speed, which led to the expansion of daily activities and the proportion of motorized travel. This brings enormous pressure on the limited urban road infrastructure and caused severe urban transportation problems. The Chinese government has put forward the public transport priority strategy and a series of transportation demand management policies to handle those problems. However, due to the lack of profound research of travel mode choice, these strategy and policies didn’ play expected roles. At the present time, the research on travel behavior in the Chinese context is limited in terms of range, depth, research ideas and methodologies. Moreover, the research findings couldn’t fully support the transportation strategy and policies. Those have led to the slow development of travel behavior theoretically and practically.

From the view of distinct characteristics, the travel decision behavior includes a sequential series of travel behaviors, from long-term behaviors like resident/job location choice, vehicle ownership choice to short-term behaviors like trip generation to mode choice, fitting the characteristics of sequential decision-making in behavioral decision theory. The research on travel decision behavior simulates the decision-making process more vividly and breaks the limitation of merely focusing on one travel behavior in previous research.

From the view of factors, the travel decision behavior is influenced by household, personal, transportation system, land-use, and attitude factors. Built environment is a complex of physical features like transportation system, land-use, urban form and the human activities patterns among the complex. Introducing the built environment factors into travel decision behavior research will provide new angle to understand travel decision behavior.

Nowadays, it is crucial to urban transportation development to handle the urban transportation problems during commute peak hours. Therefore, it is of great practical value to analyze commute travel decision behavior. Based on the views above, this dissertation conducts a research on the relationship between multidimensional factors and commute travel decision behavior. The hypothesis of the dissertation is household, personal and built environment factors are significantly associated with commute travel decision behavior and are able to explain commute travel decision behavior to some extent. The main content and conclusions of the dissertation include:

(1) The research introduces built environment factors and defines the household, personal and built environment factors as multidimensional factors. The research focus on three travel behaviors of commute travel decision behavior, which are household vehicle ownership behavior, mode-specific household commute trip generation behavior and commute mode choice behavior. To profoundly analyze the relationship between multidimensional factors and commute travel decision behavior, this dissertation then build a proper research methods framework , including the data collection, factors definition, modeling, model calibration and results analysis. Based on the research methods framework, this dissertation applies the methods to a developed medium-sized city, Zhongshan as a case study. According to the results of the case study, the dissertation finally comes to several policy implications for transportation demand management. The innovations of research content, method framework and application of theoretical methods to real case study enhance the theoretical and practical value of the dissertation.

(2) In the part of factors definition, the dissertation employs a GIS data-based multivariant statistical method to define three sub-dimensions of built environment factors. Firstly, the dissertation processes the raw data in GIS software to define the simple indicators. Then, factor analysis is used to come up with abstracted indicators. Finally, cluster analysis is employed to obtain the representative neighborhood types. According to the data of Zhongshan, ten simple indicators, five abstracted indicators and six neighborhood types were attained for further research.

(3)In the part of modeling, the dissertation elaborates the modeling method, feasibility and selection rules of multivariant linear regression models and logit models. Judging on the distribution of dependent variables, the dissertation chose Poisson regression modelfor the modeling of household vehicle ownership behavior and Negative Binomial regression model for the modeling of mode-specific household commute trip generation behavior. After examing the IIA features of the choice set, the dissertation chose Multinomiol Logit Model for the modeling of commute mode choice behavior.

(4)In the part ofmodelcalibration, the dissertation introduces the statistical software for calibration and the statistical characteristic analysis method. Using Stata 10.0 software, the calibration results clearly demonstrate the quantitative relationship between multidimensional factors and commute travel decision behavior. The results also prove the research hypothesis. The results of statistical characteristic analysis show that the model and independent variables are properly chosen.

(5)In the part ofresults analysis, based on the elasticity, the dissertation studies the profound impact of multidimensional factors quantitatively and qualitatively. In household factors, the income plays the most important role, directly influencing the household vehicle ownership behavior and indirectly influencing the mode-specific household commute trip generation behavior and commute mode choice behavior. In personal factors, the income and motorized features play the most important role, directly influencing the commute mode choice behavior. Considering the future development trends of household factors and personal factors, the commute travel decision behavior will be dominated by the motorized feature. The non-morotized vehicle ownership will be replaced by morotized vehicle ownership. Some of thenon-morotized mode and bus commute trip generation will turn into morotized mode commute trip generation. The probability of choosing morotized modes will be higher than the probability of choosing non-morotized modes and bus in commute trips.

(6) In the dimension of built environment factors, high-dense mixed land-use development and convenient destination accessibility will promote the choice ofnon-morotized modes and bus and restrain thethe choice ofmorotized modesin commute travel decision behavior, including higher ownership of non-morotized vehicles, more non-morotized and bus commute trip generation and higher probability of choosing non-morotized modes and bus in commute trips. Convenient bus service willpromote the choice ofbus and restrain thethe choice ofbicycle, moped or motorcycle in commute travel decision behavior. Well-connected road network willpromote the choice ofmorotized modes and restrain thethe choice ofnon-morotized modes and bus in commute travel decision behavior. From the aspect of neighborhood types, two urban neighborhood types are positively associated with the commute travel decision behavior in terms of walking and bicycling.Four urban and suburban neighborhood types are positively associated with the household ownership and trip generation behaviors of motorcycle and private car. Two rural neighborhood types are positively associated with the commute travel decision behavior of bus.

(7) In the part of practical application, the dissertation proposes three policies on the general planning level, including keeping high-density of land-use development, guiding mixed development and promoting the land-use pattern of jobs-housing balancing. On the comprehensive transportation planning level, the dissertation proposes two policies, including enhacing the design and construction of transit system and pay extra attention to the improvement of lower level road network. On the level of transportation management, control and organization, the dissertation proposes four policies, including the dynamic quota mechanism of private car ownership, enhancing the cost of private car ownership and usage and providing efficient and proper traveler information service.

At last, after the conclusion of main research findings and innovations, the dissertation points out the direction for future research.

Key Words:Multidimensional Factors, Commute Travel Decision Behavior, Built Environment, Multivariant Linear Regression Model, Logit Model

朱伟权

入学时间：2008级

答辩时间：2011年

论文题目：公共汽车交通运营监管基础问题研究

中文摘要：

摘要

公交是一种重要的公用事业，公交行业存在着自然垄断、经营性与公益性之间的矛盾、信息不对称以及外部性等市场失灵的问题。这些市场失灵问题的存在为政府监管的产生和发展提供了基本的理论依据，政府应该针对这些市场失灵的问题对公交行业实施监管。国际上在公交以及其他公用事业的激励性监管实践和相关实证研究都表明，引入激励性监管机制并以科学的绩效评价作为其决策支持手段对实现有效的监管具有重要意义。在我国优先发展公共交通的现实背景下，针对这些市场失灵的问题，公交监管部门要构建科学的运营监管机制和完善监管手段，提高整体监管水平，以实现对公交行业的有效监管，保障公交事业的可持续健康发展。本论文从我国公交运营监管实践出发，在交通工程学、激励性监管、绩效评价等理论和方法的指导下，重点围绕我国公交运营监管机制的构建以及支撑机制运作的公交线路的运营成本评价和管理效率评价方法展开研究，取得了一系列的研究成果。

在我国公交运营监管机制的构建研究方面，为了减轻公交行业存在的自然垄断、信息不对称问题的影响，本论文提出了在公交的成本监管以及运营绩效监管方面应引入基于标尺竞争的监管机制的总体思路并构建了总体框架。然后，一方面，根据公交企业的经营性亏损的影响因素差异，将经营性亏损划分为公交企业自身的管理因素和短期内不可控的外部运营环境因素而造成的两部分。在此基础上，为了减轻经营性与公益性之间的矛盾、信息不对称问题的影响，以及考虑到公交线路的外部运营环境的差异性，提出了完善基于成本监管的补贴机制的思路。此外，对于公交企业的主要直接运营成本的补贴，构建了应用前沿分析方法进行的公交线路运营成本评价与激励性补贴为主要手段，以实现正向激励与合理分配补贴为目的的基于成本监管的补贴机制；另一方面，为了减轻信息不对称、外部性问题的影响，构建了以公交行风考评、应用前沿分析方法进行的管理效率评价及有效奖惩为主要手段，以实现正向激励为目的的基于运营绩效评价的奖惩机制。

在公交线路的运营成本评价方法研究方面，考虑公交服务的异质性产出特性和外部运营环境的影响，基于随机前沿成本函数方法构建了公交线路的运营成本评价模型，并从公交线路的微观层面引入了四个表征其外部运营环境的指标，构建了公交线路的运营成本的评价指标体系。结合案例应用该模型对公交线路的运营成本进行了评价，获得了随机前沿成本函数和成本无效率模型的参数估计结果，

还获得了公交线路的成本效率值和可能达到的最小总运营成本等评价结果。

在公交线路的管理效率评价方法研究方面，考虑公交服务的异质性产出特性构建了对公交线路的运营效率和质量效率进行独立评价的模型框架，再基于能考虑外部运营环境影响的三阶段方法，提出了公交线路的管理效率评价的基本方法，并构建了公交线路的管理效率的评价指标体系。结合案例应用该模型对公交线路的管理效率进行了评价，获得了公交线路的运营效率值和质量效率值，以及外部运营环境因素分别对运营效率和质量效率的影响等评价结果。

本论文的研究成果在理论层面拓展了公交的激励性运营监管以及公交线路的运营成本评价和管理效率评价方法研究；在应用层面为我国公交运营监管部门构建科学的运营监管机制和完善监管手段提供了方法指导。由于公交运营监管研究的复杂性以及本人在知识结构和能力方面的欠缺，本论文还有许多不足之处有待于进一步的研究。

关键词：公交，公交线路，激励性监管，监管机制，标尺竞争，外部运营环境，运营成本，管理效率

英文摘要

ABSTRACT

Bus serviceisone of the important public utilities, which includes manyissues of market failure, suchas natural monopoly, the conflict between for-profits and non-profits,information asymmetry, and externalities and so on. These issues can be considered as the basic arguments for the motivation and evolution of government regulation. The government should regulate the bus service based on these issues of market failure. Both the practice of incentive regulation and the empirical research on regulation of international bus and other public utilities have indicated that it is significant for the efficient regulation to introduce the incentive regulatory mechanisms and scientific performance measurement as a decision making tool. Under the background of giving priority to public transport development in China, considering theseissues of market failure,the government regulator of bus need to establish the scientific regulatory mechanisms and improve the regulatory instruments, as well as increase the integral level of regulation. The purpose is to ensure the efficient regulation and the sustainable development of bus industry. The dissertation research focused on the establishment of regulatory mechanisms of bus, the measurement methodologies of operating costs and managerial efficiency of bus route, which based on the practice of bus operating regulation in China, and according to the theories and methodologies on traffic engineering, incentive regulation, and performance measurement and so on. At the same time, a series of research achievements have been obtained.

In the aspect of the establishment of regulatory mechanisms of bus in China, the dissertation proposed the regulatory mechanisms based on yardstick competition should be introduced in the costs and operating performance regulation, which in order to lighten the influences of natural monopolyand information asymmetry.

Accordingly, a general framework of the regulatory mechanisms was established. Then on the one hand, the operating deficit could be divided into the following two categories which according to the different impact factors: caused by the management factors of bus agency and by the external operating environment factors which are outside the direct control of agency in the short run.Onthisbasis,in order to lighten the influences of conflict between for-profits and non-profits,and information asymmetry, as well as considering differences of the external operating environment factors of bus route, an opinion of bettering the subsidy mechanism based on costs regulation wassuggested. Additionally, for the subsidy of main direct operating costs of bus agency, the subsidy mechanism based on costs regulation was developed. The measurement of operating costs of bus route which applies frontier analysis and incentive subsidy areservedas its main instruments. The purpose of the subsidy mechanism is to achieve active incentive and reasonable subsidy allocation. On the other hand, in order to lighten the influences of information asymmetry and externalities, a reward and punishment mechanism based on the operatingperformance measurement was designed. The measurement of bus service normalization andmanagerial efficiencywhich appliesfrontieranalysis,as well as theefficientreward and punishment areservedas its maininstruments. The purpose of the reward and punishment mechanism is to achieve active incentive.

With regard to the measurement methodology research of operating costs of bus route,considering theheterogeneous outputs of bus service and the influences of external operating environment factors, a measurement model of operating costs of bus route was established based on the stochastic frontier cost function. In addition, four new operating environment factors at the microscopic level of bus route were introduced, and a set of performance measurement indicators of operating costs of bus route was established. Then, the model was applied in theempirical research, the parameter estimate results of stochastic frontier cost function andcost inefficiency model wereacquired, and the cost efficiency scores and minimum attainable total operating costs, etc. were also obtained.

With regard to the measurement methodology research ofmanagerial efficiencyof bus route, a model framework for the independent measurement of operating efficiency and quality efficiency of bus route was developed, whichconsidering theheterogeneous outputs of bus service. Next, the basic measurement methodology of themanagerial efficiencyof bus route were proposedaccording to the three-stageanalysisapproachwhich can consider the influence of external operating environment, and a set of performance measurement indicators ofmanagerial efficiencyof bus route was established. Then, the model was applied in theempirical research, the operating efficiency and quality efficiency scoreswereacquired, and the measurement results of the effects of the external operating environment on operating efficiency and quality efficiency, etc.wereobtained.

The research achievements of the dissertation have doublesignificance.