中国安全科学学报 ›› 2023, Vol. 33 ›› Issue (9): 157-163.doi: 10.16265/j.cnki.issn1003-3033.2023.09.1676

• 安全工程技术 • 上一篇    下一篇

基于GPS数据的危险品运输车辆行程链聚类研究

陈冉冉1(), 徐佳力2, 李健1,**()   

  1. 1 同济大学 道路与交通工程教育部重点实验室,上海 201804
    2 柳州市云上龙城大数据产业发展有限公司,广西 柳州 545001
  • 收稿日期:2023-03-10 修回日期:2023-06-21 出版日期:2023-09-28
  • 通讯作者:
    **李 健(1983—),男,河南项城人,博士,副教授,主要从事应急交通规划与管理、交通大数据分析、交通政策等方面的研究。E-mail:
  • 作者简介:

    陈冉冉 (1998—),女,湖北宜昌人,博士研究生,研究方向为城市货运与应急物流。E-mail:

  • 基金资助:
    国家重点研发计划项目(2018YFB1601100); 上海市科学技术委员会科研计划项目(19DZ1202100)

Study on trip chain clustering of hazardous materials transportation vehicle based on GPS data

CHEN Ranran1(), XU Jiali2, LI Jian1,**()   

  1. 1 Key Laboratory of Road and Traffic Engineering, Ministry of Education, Tongji University, Shanghai 201804, China
    2 Liuzhou Yunshang Longcheng Big Data Industry Development Co., Ltd., Liuzhou Guangxi 545001, China
  • Received:2023-03-10 Revised:2023-06-21 Published:2023-09-28

摘要:

为帮助政府相关部门精细化监管危险化学品运输,提出一种基于全球定位系统(GPS)数据的危险品运输车辆行程链聚类分析框架。设置时空阈值,从GPS数据中提取车辆基站和中途有效停车点,生成危险品运输车辆行程链;在计算行程链数、平均停车点数、平均行程链停车时长和平均行程链距离等已有指标的基础上,使用平均行程风险和平均停车点风险衡量危险品运输车辆行程风险;以行程链特征为聚类指标,轮廓系数与误差平方和为评估指标,比选K-Means、K-Means++、PAM、FCM这4种算法的行程链聚类效果;通过实际案例分析验证框架的可行性。结果表明:K-Means++ 算法效果最佳,危险品运输车辆被分为化工园区接驳(44.19%)、城际运输(31.42%)、城市配送(13.23%)、场站接驳(9.76%)和非工作状态(1.40%)5类。

关键词: 全球定位系统(GPS)数据, 危险品运输车辆, 行程链, 聚类, 运输风险

Abstract:

In order to assist the government in regulating the transportation of hazardous materials, a cluster analysis framework based on GPS data was proposed. We set the spatiotemporal threshold, extracted the vehicle base station and halfway effective stop point from the GPS data, and generated trip chains of the hazardous material vehicle. Based on calculating the number of trip chains, the average number of stop points, the average stop time of trip chains and the average distance of trip chains, the average risk of trip and trip point were proposed to measure the risk of hazardous materials transportation vehicles. The characteristics of vehicle trip chain were taken as clustering indicators, silhouette coefficient and the sum of squares due to error were taken as evaluation indicators, and by comparing the results of K-Means, K-Means++, PAM and FCM algorithms, the trip chains of hazardous materials transportation vehicles were clustered. The feasibility of the framework was verified by case analysis. The results show that the K-Means++ algorithm has the best effect, and the hazardous material transportation vehicles are divided into 5 categories: chemical park connection (44.19%), intercity transportation (31.42%), city distribution (13.23%), freight terminal connection (9.76%) and non-working state (1.40%).

Key words: global positioning system (GPS) data, hazardous materials transportation vehicle, trip chain, cluster, transportation risk