期刊
IEEE ACCESS
卷 7, 期 -, 页码 138624-138636出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/ACCESS.2019.2943598
关键词
Metro; OD; anomaly detection; smart card; big data
资金
- Key Project of National Natural Science Foundation of China [51638004]
- Basic Research Program of Science and Technology Commission Foundation of Jiangsu Province [BK20180775]
Urban metro alleviates traffic pressure and also faces safety management problems. The metro AFC (Automatic Fare Collection System) records the OD (Origin-Destination) data of passengers' daily trips. Many researches often neglect the pretreatment of data cleaning based on smart card data. Anomaly OD records also reflect the safety problems. How to use OD to identify anomalous data and passengers' anomalous behavior is a research hotspot of metro big data. OD data of Nanjing metro were analyzed, and standard data cleaning processes were proposed including inbound records until the day before yesterday, inbound records of next days, negative records and overtime records. Then, using the data after cleaning, we analyze long-time records, short-time records, inbound and outbound records between the same stations, the swiping card records of more times, and carry out analysis. One day is chosen as an example to illustrate the analysis process, and then the OD records of several days are compared to summarize the classification of OD anomalies. Through analysis, OD anomalies can be classified into two categories: system anomalies and passenger behavior anomalies. System anomalies can be eliminated by upgrading. Abnormal passenger behavior reflects some potential safety problems. This research can effectively identify the abnormal behavior of passengers by tracking and comparing the appearing frequency of passenger cards. OD anomaly classification can be further refined, so that it has more practical value, can improve the level of metro safety management.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据