Abstract: Compared with the traditional attribute data, the traffic flow data can characterize the socio-economic connections between cities more directly and objectively. And traffic flow is a kind of data flow commonly used to reflect city network. This paper selects 88 counties and cities in Xinjiang based on highway passenger flows, and describes the pattern and characteristics of Xinjiang's urban network from the perspectives of connection strength and network node characteristics and then uses modularity class to segment the urban networks and identify its internal correlation structures. The results show that:1) The highway passenger transport connection is mainly near area connection, and the connection intensity shows obvious grade characteristics.2) The extreme characteristics of urban network nodes in Xinjiang are significantly unbalanced, and the spatial structure is quite different. The high-level node cities present the characteristics of distribution along the main highway, especially the main traffic distribution along the northern slope of Tianshan Mountain. The spatial agglomeration capacity of northern Xinjiang is higher than that of southern Xinjiang, and Urumqi is the strongest. 3) The Xinjiang urban network is divided into 10 "city groups", which have administrative boundary effects, group effects, and big city spillover effects. In summary, the overall characteristics of Xinjiang's urban network show a situation of "one core and five centers" with Urumqi as the core and Yining, Korla, Hotan, Kashgar, and Kuitun as the regional centers, which is mainly affected by the superposition of the city level and geographical distance in space.

Key words: passenger flow, city linkage, urban network, Xinjiang