Abstract: Traffic networks consist of flow system, road system, and management system. From the point of complex networks, considering the road in traffic networks as link and the influx of links as node, we have measured the danger level of links. However, the methods mentioned above can only identify road black spots after traffic accidents happened and delay the time of identifying. In this paper, an improved identification model merely based on the structure of link is proposed. Our model can estimate potential danger level of link by computing the link structure data. We select a section of link from Tanmulin to Zilaishuichuang as the research object and analyze its structure and danger level. In addition, we verify the results with equivalent accident method and physics method. As results, the model can identify danger links only by the structure data of link efficiently