基于聚类算法的车辆轨迹时空相似性研究
发布时间：2018-10-26 08:32
【摘要】：海量数据伴随新的通信技术、采集技术等的不断更新换代而不断涌现。在指数膨胀的数据中进行数据的挖掘和分析愈发重要。并且,对指导工业、农业、商业、医疗、人类行为、社会活动具有重要意义。本文中以北京市出租所车载GPS采集的时空轨迹数据为例,以数据挖掘,以及数据挖掘的方法之一——聚类为分析工具,加以数据挖掘提取相关信息。与传统的静态数据分析挖掘不相同的是,本文还加上动态时间的数据挖掘聚类,有效的得到有意义的相似性模式以及语义,再将语义模式的应用于生产生活中。首先,从数据挖掘概念以及理论和相关的聚类算法论述研究入手,掌握数据挖掘的意义及处理方法,深入了解相关算法步骤优缺点,为实际应用的时空轨迹数据的分析和挖掘提供分析技术的支撑。然后,典型的时空轨迹数据——车辆GPS采集的轨迹数据参考数据挖掘的步骤,对其预处理降噪、时间空间并行聚类、提取语义信息应用在实际问题。本课题是对将来大量多维度的数据的分析和挖掘进行探索,以及相关的数据挖掘以及聚类分析理论的概述。
[Abstract]:With the new communication technology, acquisition technology and so on, mass data are constantly emerging. It is more and more important to mine and analyze data in exponentially inflated data. Moreover, it is of great significance to guide industry, agriculture, commerce, medicine, human behavior and social activities. In this paper, the space-time track data collected by GPS in Beijing Rental Institute is taken as an example, and the data mining and clustering, one of the methods of data mining, are used to extract the relevant information. Different from the traditional static data mining, this paper also adds the dynamic time data mining clustering to effectively obtain meaningful similarity patterns and semantics, and then apply semantic patterns to production and life. First of all, starting with the concept of data mining, the theory and the related clustering algorithms, the significance and processing methods of data mining are grasped, and the advantages and disadvantages of the relevant algorithms are deeply understood. It provides the support of analysis technology for the analysis and mining of spatiotemporal trajectory data. Then, the typical space-time trajectory data, which is the reference data mining step of vehicle GPS, is preprocessed to reduce noise, time and space clustering, and the extraction of semantic information is applied to practical problems. This topic is to explore a large number of multidimensional data analysis and mining in the future, as well as related data mining and clustering analysis theory.
【学位授予单位】：福建师范大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U491
本文编号：2295156
[Abstract]:With the new communication technology, acquisition technology and so on, mass data are constantly emerging. It is more and more important to mine and analyze data in exponentially inflated data. Moreover, it is of great significance to guide industry, agriculture, commerce, medicine, human behavior and social activities. In this paper, the space-time track data collected by GPS in Beijing Rental Institute is taken as an example, and the data mining and clustering, one of the methods of data mining, are used to extract the relevant information. Different from the traditional static data mining, this paper also adds the dynamic time data mining clustering to effectively obtain meaningful similarity patterns and semantics, and then apply semantic patterns to production and life. First of all, starting with the concept of data mining, the theory and the related clustering algorithms, the significance and processing methods of data mining are grasped, and the advantages and disadvantages of the relevant algorithms are deeply understood. It provides the support of analysis technology for the analysis and mining of spatiotemporal trajectory data. Then, the typical space-time trajectory data, which is the reference data mining step of vehicle GPS, is preprocessed to reduce noise, time and space clustering, and the extraction of semantic information is applied to practical problems. This topic is to explore a large number of multidimensional data analysis and mining in the future, as well as related data mining and clustering analysis theory.
【学位授予单位】：福建师范大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U491
【参考文献】
相关期刊论文 前3条
1 蒋益娟;李响;李小杰;孙靖;;利用车辆轨迹数据提取道路网络的几何特征与精度分析[J];地球信息科学学报;2012年02期
2 荣秋生,颜君彪,郭国强;基于DBSCAN聚类算法的研究与实现[J];计算机应用;2004年04期
3 陈湘涛;李明亮;陈玉娟;;基于时间序列相似性聚类的应用研究综述[J];计算机工程与设计;2010年03期
相关硕士学位论文 前3条
1 刘爽;基于Web的数据挖掘技术研究与应用[D];长春理工大学;2007年
2 彭丽;数据挖掘中几种划分聚类算法的比较及改进[D];大连理工大学;2008年
3 苗苗苗;数据挖掘中海量数据处理算法的研究与实现[D];西安建筑科技大学;2012年
，本文编号：2295156
本文链接：https://www.wllwen.com/kejilunwen/daoluqiaoliang/2295156.html
