货车轨迹数据在公路货运系统中应用研究综述

随着轨迹数据可获取性及精度的持续提高，货车轨迹数据被广泛应用于公路货运系统的 规划与管理中，同时，人工智能和大数据分析技术的快速发展也为公路货运系统研究带来新的机 遇与挑战。本文全面梳理并总结了公路货运轨迹数据应用领域的相关研究，从基于轨迹数据的 货运出行信息辨识、货运系统关键特征预测、货运轨迹数据进一步应用3个方面回顾现有文献的 研究目标、主要内容和研究方法。通过文献分析发现：货运出行信息辨识研究聚焦于货运停留 点、车辆和货物、活动出行模式等热点主题，但现有辨识方法多移植于旅客出行研究，需要更多地 考虑货运出行的独特特征。在货运系统关键特征预测方面，研究者主要针对货运行程时间、空间 位置、出行需求等主题展开研究，并证明了基于轨迹数据预测货运特征的可行性，但预测时空范 围较为局限，需要根据具体的货运任务、货车司机特征和货运政策进行深入研究。此外，轨迹数 据也被应用于货运出行路径选择行为、货运停车休息行为、行驶安全、货运排放和能耗分析、货运 政策评估等研究。最后，在总结现有研究不足的基础上，本文认为未来研究应重点将货运轨迹数 据与其他多源数据相结合，从3个关键技术进行突破：一是针对货运实践个体，重点探索高效货车 驾驶员的出行特征和出行模式，并在货运系统中进行推广应用；二是针对交通运输新技术和新形 势，重点开发和优化自动驾驶技术和重大应急事件影响下的货运组织模式与策略；三是针对货运 供需关系及匹配机制，重点研究货运全流程供需状态辨识与预测，并结合深度学习等方法训练和 开发智能供需匹配模型，从而优化货运系统调度，助力社会散乱运力资源整合，提高货运系统的 综合效率。

Review on Application of Truck Trajectory Data in Highway Freight System

With the continuous improvement of the accessibility and accuracy of trajectory tracking data, truck trajectory data has been widely used in the planning and management of highway freight system. At the same time, the rapid development of artificial intelligence and big data analysis technology also brings new opportunities and challenges to the study of highway freight system. This paper comprehensively summarizes the researches on the application of highway freight trajectory data, and reviews the research objectives, main contents and research methods of existing literatures from three aspects: identification of freight travel information, prediction of key features of freight system, and further application of freight trajectory data. Literature analysis shows that the research on freighttravel information identification focuses on hot topics such as freight stop points, vehicles and goods, and activity travel patterns. However, the existing identification methods are mostly transplanted from the research of passenger travel, and more considerations need to be given to the unique characteristics of freight travel. In terms of forecasting the key features of the freight system, researchers mainly conduct research on topics such as freight travel time, spatial location, and travel demand, and proved the feasibility of forecasting freight characteristics based on trajectory data. However, the spatial and temporal range of prediction is relatively limited, further research is needed on specific freight tasks, characteristics of truck drivers, and freight policies. In addition, trajectory data are also further applied to freight travel route choice behavior, freight parking and rest behavior, driving safety, freight emissions and energy consumption analysis, freight policy evaluation. On the basis of analyzing the shortcomings of the existing research, this paper suggests that future research should focus on combining freight trajectory data with other multi-source data, make breakthroughs in three key technologies. First, in view of individuals of freight practice, it is necessary to focus on exploring the travel characteristics and travel patterns of efficient truck drivers and apply them in the freight system. Second, in view of new transportation technologies and new situations, the development and optimization of freight organization models, and strategies under the influence of autonomous driving technology and major emergency events should be focused. Thirdly, in view of freight supply and demand relationship and matching mechanism, the research on freight supply and demand status identification, and prediction of the whole freight process should be focused on, and the intelligent supply and demand matching model combined with deep learning methods should be trained and developed, so as to optimize the freight system scheduling, facilitate the integration of social scattered transportation resources and improve the overall efficiency of the freight system.