The network generated by urban traffic flow, also known as spatial interaction network, is an important factor in geography, which can help urban planners explore the urban regional structure and human time and space behavior. Based on the GPS trajectory data of taxis, this paper uses the complex network theory to construct a traffic flow classification model with small traffic areas as the research object, so as to study the complexity and category differences of urban traffic flow network structure, including network topology and space-time. Distribution and attribute difference characteristics. The results show that the taxi passenger flow network has scale-free characteristics; its spatial distribution has agglomeration. The classification model divides the urban passenger flow into four categories. Over time, there are obvious differences among various categories; the hotspot to hotspot area has a combined flow ratio of up to 60%; in the distance decay rate, non-hotspots. The total distance attenuation coefficient of the convergence stream type to the hotspot area is 5.02, which is the highest compared with other types. This indicates that the convergent stream type travel group is most sensitive to the distance, followed by the non-hotspot to non-hotspot area random stream type. This paper uses big data to study the urban traffic travel structure, which helps to guide urban traffic planning and reveals the interaction law between urban spatial function and crowd moving characteristics.

CICTP