交通路况在时间上和空间上具有连续变化的特征,在时空维度上对交通路况进行高分辨率采样得到的数据,对研究交通路况的时空动态十分有利。但长时间大范围的高分辨率交通路况信息数据量巨大,给数据的组织和管理带来了困难。目前,尚没有一种成熟的时空数据模型对高时空分辨率交通路况数据进行高效(顾及数据存储与访问效率)的组织管理。本文提出一种基于线性参照系统的交通路况基态修正模型。此模型应用基态修正模型的基本思想,在时间维度上对交通路况数据进行无损压缩,又引入动态分段技术和线性参照系统,以路划作为交通路况载体,在空间维度上对交通路况数据进行压缩存储。利用成都市区真实交通路况数据,本文验证了此模型的有效性,比较了6种不同参数下交通路况基态修正模型的存储和访问效率,给出了最佳模型建议。

Road traffic condition constantly changes in both spatial and temporal domains. Traffic information of high spatiotemporal resolution is valuable to the study of road traffic dynamics. However, the massive data volume of high resolution traffic information of large spatiotemporal extent introduces difficulties in data organization and management. Actually, there is no well-accepted spatiotemporal data model specialized for management of such data that is effective in terms of data storage and access efficiencies. To overcome such drawbacks, this paper proposes a base state amendment model (BSAM) for dynamic traffic condition, which is based on linear referencing system. The model utilizes the basic strategy of BSAM to do lossless data compression in the time dimension while compresses data in the spatial domain by utilizing linear referencing system (LRS) and dynamic segmentation. In addition, road stroke instead of road segment is used for route identification in the LRS to further reduce data volume. We validate the effectiveness of the proposed model through its use in traffic condition data modeling in Chengdu urban area. Six variants of the proposed model and their characteristics are analyzed and compared. The six variants of the proposed model are denoted as BSAMa, BSAMb, BSAMc, BSAMd, BSAMe and BSAMf, respectively (see section 3). The main differences between them are in the number of base states, time interval between two base states and the number of non-base states. Data storage and accessing efficiencies of the six variants are qualitatively and quantitatively analyzed. The results indicate that BSAMf (see Fig. 4) has the optimal storage and accessing efficiencies for the data used. Furthermore, a method is suggested to estimate the number of non-base states between successive base states in the BSAMf model, which is based on the dynamic characteristics of the traffic condition data. With the data of Chengdu, it is demonstrated that two non-base states between successive base states is the most appropriate pattern for BSAMf in terms of data storage and accessing efficiencies. In sum, the experiment results demonstrate that the proposed BSAM for traffic condition data is effective and efficient.