Abstract: Heavy freight vehicles are not only involved in urban economic production and logistics efficiency, but also have a significant impact on the quality of urban space due to their emission, noise, congestion and other externalities. Therefore, it is an urgent need to analyze the spatiotemporal characteristics and influencing factors of their activities for the improvement of urban space quality and the refinement of traffic management. Based on the Global Positioning System (GPS) trajectory data of heavy trucks over 12 tons registered in Shenzhen and considering the operation differences of different types of freight vehicles, this paper proposes a method to determine the travel cutting threshold level of trajectory data through the observation results of the minimum sample size in statistical sampling. The discrepancies in the trip intensity and travel period are analyzed for 5 types of heavy freight vehicles, with a comparison to the passenger vehicles. The contrast manifests that their activities are staggered in time and the nighttime travel accounts for a higher proportion. Furthermore, the activity space of heavy freight vehicles is diverse within the group. Container trucks have widely cross-city activity demand and are more responsible for medium and long-distance transportation, while earthmoving vehicles and heavy tank trucks are more localized service transportation functions. The activity space of heavy freight vehicles has the characteristics of agglomeration, and the node intensity of travel activity has the characteristics of scale-free power law distribution. The generalized additive model was used to analyze the influencing factors of the activity space differences of two typical trucks. Container truck's activity shows a significant non-linear correlation with logistics facilities such as ports and storage parks, while ordinary large truck's activity is closely related to industrial parks. This paper can provide new perspectives and methods for the effective management of heavy truck traffic and heavy truck demand modelling.

Key words: urban traffic, spatiotemporal distribution, trajectory data mining, truck traffic, influencing factor, generalized additive model

摘要： 重型货运车辆能促进城市经济生产和物流效率，但也因显著的排放、噪声和拥堵等外部属性对城市空间品质有明显影响，对其活动时空特征和影响因素分析是当前城市空间品质提升和交通管理精细化的迫切需求。基于深圳注册的12 t以上重型货车GPS轨迹数据，考虑不同类型货运车辆作业差异，提出利用统计抽样中最小样本量的观察结果确定不同类型重型货车轨迹数据出行切割阈值水平的方法；针对5类重型货车，分析出车强度和出行时段的差异性，与小客车进行对比分析发现，重型货车出行活动与小客车出行在时间上存在错峰特征，且夜间出行比例较高；重型货车的活动空间具有组内差异性，集装箱卡车具有显著的跨城市活动需求，更多承担中长距离运输功能，而土方车和重型罐式车更多承担本地化运输服务功能；重型货车的活动空间具有集聚特征，出行活动的节点强度具有无标度幂律分布特征；利用广义加性模型分析两类典型货车活动空间的影响因素，集装箱卡车活动与港口及交通仓储园区等物流设施呈现非线性关系，而普通大货车与工业园区等产业用地具有紧密关系。本文可以为重型货车交通精细化管理和交通需求模型建设提供新的思路和方法。

关键词: 城市交通, 时空分布, 轨迹数据挖掘, 货车交通, 影响因素, 广义加性模型