基于浮动车数据的居民出行行为的动力学模型及特征分析

发布时间：2018-02-04 08:40

本文关键词： 复杂系统排队论模型动力特征分析出行行为　出处：《哈尔滨工业大学》2015年硕士论文　论文类型：学位论文

【摘要】：近年来,随着社会的发展和人类对社会服务领域要求的提高,居民出行行为的特征与方式逐渐成为了一个热门的课题,掌握更多的居民出行行为的信息就能够更好的服务于人类的生活。长久以来,由于居民出行行为在其模型建立和应用特征的分析上具有双重价值,吸引了许多领域学者的共同关注。本文研究的主要内容是:在居民出行行为的动力学模型方面,首先针对个体和群体模型进行研究,之后分别利用模拟数据和实证数据验证其准确性,最后再产生替代数据并验证其准确性;在居民出行行为的特征分析方面,将浮动车数据可视化到地图,融合聚类分析算法,挖掘出深圳市居民上下车的兴趣点。此外,我们以此为基础针对深圳市出租车的载客模式进行了特征提取,挖掘典型的载客特征。论文的主要内容包括下面几个部分:在模型研究方面,本文对个体和群体出行分别进行研究。针对个体出行,从出租车和乘客两者角度出发,以载客的时间间隔为切入点,分别推导有关于出租车载客次数和乘客打车次数的模型。针对群体出行,基于对排队论模型的理解,对原有的人类动力学模型进行推广改进,推广改进模型后分别应用模拟数据以及实际的数据对其进行验证。该模型分别选取了具有不同特性的个体去推导群体的特性,但群体的结果却都服从幂律分布,从而能够说明群体出行的特征并不是个体出行特征的叠加。在验证改进模型的准确性后,基于改进模型产生居民出行行为的替代数据,并验证其准确性,从而丰富了用来研究人类行为动力学的数据。在特征分析方面,选择合适的地图匹配算法,完成浮动车数据的地图可视化。之后,基于对浮动车数据的统计分析,使用直观形象的地图可视化工具完成居民出行行为的兴趣点挖掘,生成热应力图,并对居民出行行为进行特征分析。为了验证使用上述地图可视化工具挖掘居民出行兴趣点的准确性,使用更加精确的聚类分析算法对浮动车数据进行处理。本文通过比较不同的聚类分析算法,选择凝聚式层次聚类分析算法对浮动车数据进行挖掘,并结合更加精确的电子地图进行地图可视化,通过与热应力图所呈现出来的兴趣点比对,验证其准确性。基于验证准确的热应力图为基础,从出租车的运行区域出发对深圳市出租车的载客模式进行归纳总结,我们认为居民在出行的空间位置上近似服从于正态分布,由此本文完成了对居民出行行为的时间间隔和空间跨度的模型建立和特征分析。
[Abstract]:In recent years, with the development of society and the improvement of human requirements in the field of social services, the characteristics and patterns of residents' travel behavior have gradually become a hot topic. Having more information on the travel behavior of residents can better serve the human life. For a long time, because of the resident travel behavior in its model building and application of the analysis of the characteristics of the dual value. It has attracted the common attention of many scholars. The main content of this paper is: in the dynamic model of resident travel behavior, the individual and group models are studied first. After that, the simulation data and empirical data are used to verify its accuracy, and then substitute data are generated to verify its accuracy. In the aspect of characteristic analysis of residents' travel behavior, the floating vehicle data is visualized to the map, and the clustering analysis algorithm is combined to find out the points of interest of the residents in Shenzhen. Based on this, we extract the features of the passenger mode of Shenzhen taxi, mining the typical passenger characteristics. The main content of this paper includes the following parts: in the research of the model. In this paper, the individual and group travel are studied separately. For individual travel, from the perspective of taxi and passenger, take the time interval of passengers as the breakthrough point. For group travel, based on the understanding of queuing theory model, the existing human dynamics model is extended and improved. The improved model is validated by simulation data and actual data. The model selects individuals with different characteristics to deduce the characteristics of the population, but the results of the population are distributed according to the power law. It can show that the characteristics of group travel is not the superposition of individual travel characteristics. After verifying the accuracy of the improved model, the improved model can generate the replacement data of the residents' travel behavior, and verify its accuracy. It enriches the data used to study the dynamics of human behavior. In the aspect of feature analysis, the appropriate map matching algorithm is selected to complete the map visualization of floating vehicle data. Based on the statistical analysis of floating vehicle data, the visual map visualization tool is used to mine the points of interest of residents' travel behavior and generate the thermal stress map. In order to verify the accuracy of using the above map visualization tools to mine the residents' travel interest points. Using more accurate clustering analysis algorithm to deal with floating vehicle data. By comparing different clustering analysis algorithms, this paper selects the condensed hierarchical clustering analysis algorithm to mine floating vehicle data. Combined with the more accurate electronic map to visualize the map, the accuracy of the map is verified by comparing it with the points of interest shown in the thermal stress map, which is based on the verification of the accurate thermal stress map. Starting from the operating area of taxis in Shenzhen, we summarize the passenger carrying mode of taxis in Shenzhen. We think that the residents in the travel space position is similar to the normal distribution. In this paper, the time interval and spatial span of residents' travel behavior are modeled and analyzed.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U491

【参考文献】