GPS轨迹信息的语义挖掘

发布时间：2018-03-03 22:35

本文选题：出行调查　切入点：GPS轨迹数据　出处：《山东理工大学》2013年硕士论文　论文类型：学位论文

【摘要】：近年来城市化进程加剧,在较短时间内城市人口急剧增长,这考验着城市各方面的承载能力,尤其对城市交通提出了更高的要求。交通调查是交通理论研究和技术创新的基石,其中居民出行信息更是重要的调查内容。目前广泛应用的居民出行调查法存在周期长、成本高、数据质量不高等问题,已逐渐不能满足大规模、高频率的居民出行调查的要求。伴随无线通信网络和全球定位系统(GPS)技术的迅猛发展,海量GPS数据的收集和传输成为可能,基于GPS的出行调查法应运而生。该方法是指给受访者配备一个GPS接收器,采集其出行的轨迹数据,通过数据挖掘及语义挖掘技术,智能化的提取数据中所隐含的居民出行信息。本文对基于GPS的出行调查法的研究,围绕从无直接意义的数据中智能提取出行信息。主要提取行程、出行方式和出行目的三类信息,具体如下： (1)行程识别行程识别是出行信息提取的首要步骤。文中通过基于密度的轨迹点聚类获取轨迹的低速区域,也就是受访者可能的停留位置；将低速区域匹配到GIS地理信息系统上,进一步判断低速区域是否为停留。辨识出轨迹中的停留,即找到了行程端点,也就完成了行程识别的过程。 (2)基于模糊模式识别的出行方式判别出行方式是出行信息提取的重点。文中针对出行方式模糊性的特点,提出使用模糊模式识别进行出行方式判别。利用主成分分析法确定出特征变量,用以表征行程段出行方式信息；对应步行、自行车和机动车这三种出行方式分别建立隶属函数,用matlab实现模糊模式识别模型的构建,使用模型进行出行方式判别。 (3)基于多级空间尺度的出行目的推断出行目的是出行信息提取的难点。文中利用地理学中多级空间尺度理论,在不同级空间中分析GPS轨迹。着重剖析轨迹的微观活动,从轨迹停留中进一步辨识子停留。挖掘子停留的语义信息,用轨迹点特征参数(时长、速度、转角)对信息进行量化。在大量数据统计结果基础上构建判别信息库,将子停留信息与判别信息库中阀值进行比对,得知子停留活动类型,继而获知出行者的出行目的。
[Abstract]:In recent years, the urbanization process has intensified and the urban population has increased rapidly in a relatively short period of time, which tests the carrying capacity of various aspects of the city, especially puts forward higher requirements for urban traffic. Traffic survey is the cornerstone of traffic theory research and technological innovation. Among them, the resident travel information is an important investigation content. At present, the widely used resident travel survey method has many problems, such as long period, high cost, low data quality and so on, which can not meet the requirements of large-scale and high-frequency residents' travel survey. With the rapid development of wireless communication network and GPS (Global Positioning system) technology, it is possible to collect and transmit huge amounts of GPS data, and the GPS based travel survey method comes into being, which means that the interviewees are equipped with a GPS receiver. Through data mining and semantic mining technology, we can intelligently extract the resident travel information implied in the data. In this paper, we study the trip survey method based on GPS. The travel information is extracted intelligently from the data without direct meaning. There are three kinds of information: itinerary, travel mode and travel purpose. The details are as follows:. Stroke identification. Travel identification is the first step to extract travel information. In this paper, the low speed region of trajectory is obtained by density-based locus clustering, that is, the possible stay position of interviewee; the low speed region is matched to GIS GIS. Furthermore, it is determined whether the low speed region is a stopover. The identification of the stopover in the trajectory, that is to say, finding the end point of the stroke, will also complete the process of the travel identification. Identification of trip modes based on Fuzzy pattern recognition. Trip mode is the key point of trip information extraction. In view of the fuzziness of trip mode, fuzzy pattern recognition is used to distinguish trip mode, and principal component analysis is used to determine the characteristic variable. It is used to represent travel mode information of travel segment. Membership function is established for three travel modes namely walking bicycle and motor vehicle. Fuzzy pattern recognition model is constructed by matlab and trip mode identification is carried out by using the model. Travel destination inference based on multilevel spatial scale. The purpose of travel is difficult to extract travel information. In this paper, we use the theory of multilevel spatial scale in geography to analyze the GPS locus in different levels of space. The information is quantified by the characteristic parameters of the locus points (time, speed, angle), and the discriminant information base is constructed on the basis of a large number of statistical results. The sub-stay information is compared with the threshold value in the discriminant information base, and the type of sub-stay activity is obtained, and then the travel purpose of the traveller is obtained.
【学位授予单位】：山东理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P228.4;U495

【参考文献】