摘要: |
The shipments of hazardous materials (hazmat) which are indispensable for economic and social development have increased; accordingly, a rising number of incidents involving hazmat transportation may inflict more dread damages to both people and environment. This severe situation has prompted the need for deep mining trip purposes using trajectory information in order to enhanee the hazmat-transportation regulatory. This paper presents an unsupervised two-phase framework for inferring multiple trip purposes (i.e. loading, unloading, in-yard, and other stops) based on the passive global positioning system (GPS) data during the hazmat-transportation process. In detail, a scalable ordering points to identify the clustering-structure mixture algorithm (SOMA) is first developed to group hazmat vehicles trip ends into hotspot places in phase I; In phase II, a two-stage trip-purpose identl-fication approach is proposed with a combination of the fuzzy c-means (FCM) method and the point-of-interest (POI) information. The effectiveness and efficiency of the designed two-phase framework are evaluated through the real-world datasets, which are gen erated by more tha n 12,000 vehicles in Liaoning Provi nee, China. The results dem on strate that the method can infer four types of freight trip purposes with an accuracy of 82.1%. The proposed approach framework can help analyze the vehicle trips associated with the loading states, which will provide effective decision-making support for the hazmat-transporta-tion regulatory. |