【作者】 徐弘祥；

【导师】 侯立文；

【作者基本信息】 上海交通大学 ， 管理科学与工程， 2011， 硕士

【摘要】 在梳理了国内外学者研究成果的基础上,提出了现有动态共乘研究的不足之处,从运行规则和匹配效率上对动态共乘问题进行了探索,研究如何在有时间窗约束的情况下匹配个体的供给与需求。建立了“单司机单乘客”的基本动态共乘模型以及相应的规则,在此基础上拓展了个体基于偏好进行共乘选择的条件和方法。除此之外,还探索了单司机多乘客的复杂共乘模型,提出了基本的模型假设和运行规则。文中设计了两种匹配算法,并利用Netlogo平台开发了仿真程序,通过系统仿真的方式获得了大量动态共乘系统模拟运行的数据,以此评价时间窗长度、个体到达速率、车辆行驶速度、以及司机的期望比例等参数对系统运行指标的影响。这些指标包括系统平均匹配率、全局时间节省比率、个体平均等待时间等等。文章主要研究的问题包括“系统运行结果的主要影响因素”、“不同匹配方式之间的相互比较”、“偏好选择条件对系统运行结果的影响”、以及“系统运行指标在不同系统参数的组合下的变化趋势”。从仿真结果可以看到,时间窗对动态共乘的结果有着重要的影响。其他系统参数对于最后的结果也都有不同程度的影响。文章通过回归分析对其显著性进行了验证。在不同匹配方式的比较中,使用方差分析验证了不同的匹配方式对于各运行结果是否具有显著差异。更多还原

【Abstract】 Based on the analysis of researches at home and abroad, this thesis presents the limitations of current studies of dynamic ridesharing and explores the problem in the aspects of operational rules and match-making efficiency to figure out how to match the needs and demands of individuals under the constrain of time windows. Besides, the thesis establishs a“one-driver-one-rider”dynamic ridesharing model with its relevant rules and expands the ridesharing selection methods and conditions for individuals on the basis of preference. Further more, a more complicated“one-driver-mutliple-riders”dynamic ridesharing model is explored and its basic assumptions and operational regulations are presented.The thesis designs two match-making methods and develops simulation programs via Netlogo platform. By means of system simulation, assessment is made according to the data, which reveals the impact of system parameters like time window length, user involvement, car speed and the expectional proportion of drivers on system indexes, which include match-making percentage, time-saving percentage and average waiting time of individuals.The research problems of this thesis include“the major influencing factors on the results of system operation”,“the comparison between different match-macking methods”,“the influence of preference on the result of system running”and“the trend of system indexes under different combination of system parameters”. The results of system simulation show that the length of time window has a greate impact on the performance of dynamic ridesharing, and other system parameters also influence the results in varying degrees. The thesis verifies the significance by regression analysis and ANOVA is used to inspect whether different match-making methods have significant difference in the results of system operaton.更多还原