Journal of Computer Applications ›› 2018, Vol. 38 ›› Issue (7): 1923-1928.DOI: 10.11772/j.issn.1001-9081.2017123041

Previous Articles     Next Articles

Map matching algorithm based on massive bus trajectory data mining

CHEN Hui1, JIANG Guifeng1, JIANG Guiyuan2, WU Jigang1   

  1. 1. School of Computers, Guangdong University of Technology, Guangzhou Guangdong 510006, China;
    2. School of Computer Science and Engineering, Nanyang Technological University, Singapore 639798, Singapore
  • Received:2017-12-26 Revised:2018-02-08 Online:2018-07-10 Published:2018-07-12
  • Supported by:
    This work is partially supported by the Guangdong Provincial Science and Technology Program (2017A040402009).

基于海量公交轨迹数据挖掘的地图匹配算法

陈辉1, 蒋圭峰1, 姜桂圆2, 武继刚1   

  1. 1. 广东工业大学 计算机学院, 广州 510006;
    2. 南洋理工大学 计算机科学与工程学院, 新加坡 新加坡 639798
  • 通讯作者: 蒋圭峰
  • 作者简介:陈辉(1974-),男,广东梅州人,副教授,硕士,主要研究方向:计算机网络、云计算、数据科学;蒋圭峰(1991-),男,江西上饶人,硕士研究生,主要研究方向:机器学习、数据挖掘、数据科学;姜桂圆(1985-),男,山东临沂人,博士,主要研究方向:城市智能交通系统、交通大数据处理、交通资源优化;武继刚(1963-),男,江苏沛县人,教授,博士,CCF会员,主要研究方向:数据科学、网络设计、机器智能、容错计算、软硬件协同设计。
  • 基金资助:
    广东省省级科技计划资助项目(2017A040402009)。

Abstract: Concerning poor matching effect of existing map matching algorithms (such as classical Hidden Markov and its variants, advanced algorithms) for low-frequency trajectory data, a trajectory data mining method based on massive bus historical trajectory data was proposed. Taking bus stations as the sequence skeleton firstly, by mining, extracting, regrouping and sorting trajectory data from a large number of low frequency trajectory points to form high frequency trajectory data, then the high-frequency trajectory data sequence was processed by the map matching algorithm based on classical hidden Markov model to get the bus route map matching results. Compared with the matching method on the low-frequency data not processed by the mining algorithm, the proposed method reduces the matching error by an average of 6.3%, requires smaller data size and costs less time. In addition, this method is robust to low-frequency, unstable noise trajectory data, and it is suitable for map matching of all bus routes.

Key words: bus trajectory data, map matching, data driven, high frequency trajectory data mining

摘要: 针对现有地图匹配算法(如基于经典隐马尔可夫及其变体、先进算法等)对于低频轨迹数据匹配效果不甚理想的问题,提出一种基于海量公交历史轨迹数据的轨迹数据挖掘方法。首先,以公交站点为序列骨架,从大量低频轨迹中挖掘、提取轨迹点数据,进行重组、排序形成高质量高频轨迹数据序列;然后,将高质量高频轨迹数据序列应用基于经典隐马尔可夫模型地图匹配算法,得到公交路线地图匹配结果。与未经过挖掘算法处理的低频轨迹数据的匹配方法相比,所提方法在匹配误差上平均下降6.3%,匹配所需的数据规模、时间大幅缩减;且该方法对于低频、不稳定的噪声数据具有鲁棒性,适用于所有公交路线的地图匹配问题。

关键词: 公交轨迹数据, 地图匹配, 数据驱动, 高频轨迹数据挖掘

CLC Number: