| 基于手机信令数据的出行端点识别效果评估 |
| |
| 引用本文： | 杨飞,姜海航,姚振兴,刘好德.基于手机信令数据的出行端点识别效果评估[J].西南交通大学学报,2021,56(5):928-936. |
| |
| 作者姓名： | 杨飞 姜海航 姚振兴 刘好德 |
| |
| 基金项目： | 国家重点研发计划（2018YF1600900）；国家自然科学基金（51678505）；中央高校基本科研业务费专项资金（300102219301、300102210204）；贵州省交通运输厅科研项目（2018-321-026）；教育部人文社会科学基金青年基金项目（20XJCZH011）；高等学校学科创新引智计划资助（B20035） |
| |
| 摘 要： | 为了研究利用手机信令数据识别个体出行端点的应用效果，开展实地采集手机信令数据的出行试验，且同步采集相应的GPS轨迹数据和出行日志作为算法评估的真实数据，提出出行端点识别的3阶段处理算法. 首先，提出等时距补点算法平衡各信令定位点的时间权重；然后，利用凝聚层次聚类算法将定位点聚类成不同的类簇；最后，针对已有研究中缺乏关注的类簇震荡现象，提出新的震荡修正算法对聚类结果做进一步优化. 案例结果表明:本文提出的方法对出行端点识别的精度、距离误差和时间误差上均有较好的效果，出行端点识别个数的精度在84%以上，端点位置识别距离平均误差在220 m以内，出行端点的离开和到达时间的平均误差分别为7.7 min 和5.3 min；在不同的出行目的的比较中，以工作为目的的端点识别效果最好，以娱乐购物为目的的端点识别效果相对较差.
|
| 关 键 词： | 出行调查 出行端点识别 手机信令 层次聚类 端点震荡修正算法 |
| 收稿时间： | 2020-03-11 |
Evaluation of Activity Location Recognition Using Cellular Signaling Data |
| |
| YANG Fei,JIANG Haihang,YAO Zhenxing,LIU Haode.Evaluation of Activity Location Recognition Using Cellular Signaling Data[J].Journal of Southwest Jiaotong University,2021,56(5):928-936. |
| |
| Authors: | YANG Fei JIANG Haihang YAO Zhenxing LIU Haode |
| |
| Abstract: | In order to investigate the recognition results of individuals’ activity locations using cellular signaling data, the field experiment for collecting cellular signaling data was carried out. The GPS trajectory data and the travel logs were collected synchronously as the real data for reference. A three-step method for recognizing activity locations is proposed. Firstly, an equal time interval interpolation method is used to balance the time weight of each trace. Secondly, an agglomerative hierarchical clustering algorithm is applied to merge the traces into different clusters. Finally, a new method of correcting location oscillation is proposed, to solve the problem that clusters in the same activity location oscillate. Results show that the proposed method performs well in term of the accuracy of identifying activity locations, distance error and time error. The average recognition accuracy and distance error is over 84% and within 220 m, respectively. The average errors of departure and arrival time are 7.7 min and 5.3 min, respectively. In the comparison of different travel purposes, the activity locations for work receive the best recognition results, and the results of the locations for shopping is relatively inferior. |
| |
| Keywords: | |
|
| 点击此处可从《西南交通大学学报》浏览原始摘要信息 |
|
点击此处可从《西南交通大学学报》下载全文 |
|
