Classification of the freight trip purpose of heavy trucks using trajectory data and waybill data

Heavy trucks are pivotal to large-scale freight logistics, but their operations generate substantial negative externalities, such as increased accident risks and inefficient resource use. A fundamental yet unresolved challenge in characterizing heavy truck trip patterns is the comprehensive classification of freight trip purposes through the analysis of extensive freight-related data. Overcoming this challenge is essential for enhancing safety outcomes and advancing freight system optimization. Previous studies have primarily focused on three trip types: loading, unloading, and returning to base. However, a fourth type, termed unloading-loading, where partial or full unloading is followed by reloading at the same stop, has yet to be considered despite its fundamental importance in characterizing heavy truck trip patterns. To bridge this gap, we begin by classifying freight trip purposes into these four types according to whether a truck carries commoditiesacross two consecutive trips. We then propose a rule-based model to identify freight origins, destinations and trip purposes from large-scale freight data. Recognizing limitations in waybill data, we supplement this approach with a machine learning model. Finally, we extract heavy truck trip activity chains and reveal their underlying diversity and heterogeneity. Our results have wide applications, such as the analysis of heavy truck freight transportationefficiency and the development of a freight trip distribution model.