【作者】 李明；

【导师】 杨小宝；

【作者基本信息】 北京交通大学 ， 智能交通工程， 2015， 硕士

【摘要】 随着社会经济的发展,消费需求的增加,购物出行占城市居民日常出行的比例不断增大。出行时间是智能交通系统中交通预测的重要参数之一,出行时间受到个人属性、家庭属性和出行方式等诸多因素的影响。与通勤出行相比,购物出行在时间上有着很大的灵活性,购物出行的时间分布更为复杂。因此,研究城市居民购物出行时间的分布规律及其影响因素有着重要的实际意义。本文以S省省会城市居民出行调查数据为基础,运用生存分析中的风险模型,建立购物出发时间的选择模型,识别出影响居民购物出发时间的关键因素,进而预测当这些因素发生变化时对出发时间选择的影响。研究结果可以给智能交通系统提供更加直观、精确的预测信息,为智能系统的交通诱导和管理控制提供科学依据。论文的主要工作如下：(1)购物出发时间的整体分布规律。基于居民出行调查数据,建立购物出发时间的风险模型,对购物出发时间进行估计。结果表明在上午7：00之前,生存率变化缓慢,出发去购物的人很少。从早上7：00持续到10：00,生存率急剧下降,为购物出行的高峰,出行集中且持续时间较长,出行量约占全日出行量的65.4%。16：00之后,随着时间的增加,风险率函数的趋势是上升的,说明在16：00前仍未去购物的人群,在这之后出去购物的风险率越来越大。(2)购物出发时间的Cox风险模型及其影响因素分析。构建购物出发时间的Cox风险模型,对模型参数进行估计,分析各因素对购物出发时间的影响。结果表明性别、年龄、是否有6周岁以下儿童、中等月收入、出行方式这5个因素对购物出发时间均有显著的影响。总体来说,男性购物出发时间早于女性,年龄为60岁以上的老年人购物出发的时间最早,中等收入水平的人群购物出发时间早于其他收入水平的人群,家庭中有6周岁以下成员的要比没有6周岁以下成员的家庭出发时间更早,出行方式中步行的出发时间最早。Cox-Shell残差图的检验结果表明Cox模型的拟合效果良好。并预测当出行方式和年龄的比例发生变化时,对出发时间的影响。(3)基于参数模型的活动出发时间分析。基于实证数据,运用加速失效时间模型,找出最优的参数模型,并揭示影响购物出发时间的关键因素。结果表明log-logistic模型最适合用来拟合购物出发时间的分布。性别、是否有6周岁以下的儿童、年龄、出行方式等是影响购物出发时间的重要因素。更多还原

【Abstract】 With the development of society and economy and the increase of the consumer demand, the proportion of shopping trips has growing trend. Departure time is one of the important parameters of traffic prediction in Intelligent Traffic System; It is affected by the personal attributes, family property, travel mode, and many other factors. Compared with commuting, shopping trip has great flexibility in time. The time distribution of shopping is more complex. Therefore, it has important practical significance to study the distribution and affecting factors of urban shopping trip. Based on the survey data of Jinan city resident, this paper uses hazard model in survival analysis. It establish the departure time choice model of shopping, and analysis the effects of different social economy on shopping departure time, then predict change in departure time. The research results can provide predictive information for the Intelligent Transportation System more intuitive, accurate.At the same time; it provides the scientific basis for traffic inducement and management of intelligent system. The main work is as follows:(1) The overall departure time distributions of urban shopping were explored. Based on the residents travel investigation data, a hazard model of departure time of urban shopping was proposed. Their departure time was estimated. The different risk rate of social economic variables was explored. The results showed that before7:00in morning, the survival rate changed slowly. From the morning7:00continued to10:00, survival rate decreased dramatically. About58.5%.of travelers have gone shopping. After16:00, with the time increasing, the trend of hazard rate is upward.(2)Cox hazard-based models and the influence factors of departure time were investigated. Cox hazard-based models of departure time were proposed. Based on the surveyed data, the model parameters were estimated. The significant factors on departure time were analyzed. The results show that gender, age, whether to have children less than6years of age, medium income, and travel mode have significantly influence on departure time behavior. Overall, in departure time, male are earlier than the female, above60years old start the earliest than other age, middle income people are earlier than other income levels, family with children under6years start earlier than family without children. Walking start the earliest than other travel modes. Test Cox-Shell residual plot results show that the Cox model fitting effect is good.(3)Departure time of urban shopping was proposed by using parameter theory and, accelerated hazard models method. Based on the empirical data, the optimal parameter model was chosen. And key factors affecting the departure time were investigated. The results show that the log-logistic model is the most suitable to fit departure time of urban shopping. In the analysis of influencing factors, gender and family whether with children less than6, age, travel mode are important factor.更多还原