支持智能交通的数据分析技术及其应用研发

发布时间：2018-01-11 00:13

本文关键词：支持智能交通的数据分析技术及其应用研发　出处：《南京大学》2015年硕士论文　论文类型：学位论文

【摘要】：随着经济的发展、城市化进程的不断深入,传感器技术、通信技术、地理信息系统(GIS)技术和计算机技术的不断发展,智能交通技术应运而生。它将先进的科学技术有效地综合运用于整个地面交通管理系统,加强车辆、道路、使用者三者之间的联系,全方位全天候发挥作用,从而形成一种保障安全、提高效率、改善环境、节约能源的综合运输管理系统。物联网和通信技术的快速发展使得交通信息的发布不再是瓶颈,要实现交通信息应用的持续快速增长,如何获取原始交通数据并处理成精准的交通信息是其关键。可以说,交通信息的采集和处理是智能交通系统的关键技术。目前对于实时交通数据的采集主要有两种方式:一种是静态交通探测;一种是动态交通探测。车牌识别属于静态交通探测的一种,由于它能自动快速的提取车牌信息,得到了广泛的应用。然而,受制于光照、气候、车辆速度、遮挡物体或传感器失灵等不可控因素的干扰,自动车牌识别的准确度一直受到外部或内部因素的干扰,在实际应用中经常出现识别效果不理想的情况。作为动态交通流信息采集的主要手段,GPS技术在国内外得到了广泛的应用,可以实时提供三维坐标、速度等空间信息。交通领域使用GPS最主要群体是公交和出租车,大量的GPS记录,包含了一个城市交通的客观属性和规律,也反映出出租车司机主观上的行驶习惯,提供了一个供我们观察分析出租车司机的窗口,而目前这些数据尚未得到充分应用。鉴于上述问题,本文从交通信息历史记录出发,首先提出了自动车牌纠错方法,该方法独立于自动车牌识别方法,而从一系列识别后的数据出发,尝试纠正识别错误的车牌,并给出各个设备识别错误的原因;然后就出租车GPS数据,提出了一个在线的异常轨迹检测方法,该方法分为路径推荐和异常检测两部分,首先从出租车历史轨迹数据出发,给出起点和终点之间的路径推荐,然后,在线的对出租车行驶过程进行异常检测。
[Abstract]:With the development of economy and urbanization, sensor technology, communication technology, geographic information system (GIS) technology and computer technology are developing continuously. Intelligent transportation technology emerges as the times require. It effectively applies advanced science and technology to the whole ground traffic management system, strengthens the connection among vehicles, roads and users, and plays an all-weather role in all directions. A comprehensive transportation management system is formed to ensure safety, improve efficiency, improve environment and save energy. With the rapid development of Internet of things and communication technology, the release of traffic information is no longer the bottleneck. In order to realize the continuous and rapid growth of traffic information application, how to obtain the original traffic data and process it into accurate traffic information is the key. The acquisition and processing of traffic information is the key technology of its. There are two main ways to collect real-time traffic data: one is static traffic detection; One is dynamic traffic detection. License plate recognition is a kind of static traffic detection. Because it can extract license plate information automatically and quickly, it is widely used. However, it is restricted by illumination, climate and vehicle speed. The accuracy of automatic license plate recognition is always interfered by external or internal factors due to the interference of uncontrollable factors such as occlusion object or sensor failure. As the main means of dynamic traffic flow information collection, GPS technology has been widely used at home and abroad, which can provide three-dimensional coordinates in real time. Speed and other spatial information. The most important group using GPS in the field of transportation is bus and taxi. A large number of GPS records include the objective attributes and laws of a city traffic. It also reflects the subjective driving habits of taxi drivers and provides a window for us to observe and analyze taxi drivers, and these data are not yet fully applied. Based on the history of traffic information, this paper first proposes an automatic license plate error correction method, which is independent of the automatic license plate recognition method, and from a series of data after recognition, try to correct the recognition of the wrong license plate. The causes of identifying errors for each equipment are also given. Then on the taxi GPS data, an online anomaly track detection method is proposed, which is divided into two parts: path recommendation and anomaly detection. Firstly, starting from the taxi history track data. The path recommendation between the starting point and the end point is given, and then the abnormal detection of taxi driving process is carried out online.
【学位授予单位】：南京大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495;TP311.13

【相似文献】