ABSTRACT
Traffic time series data is the data representation of traffic flow characteristics, which contains the internal mechanism and evolution of the traffic system. To analyze the characteristics of traffic flow time series from multiple angles, a complex network of time series is constructed based on phase space reconstruction theory and critical radius connection strategy. At the same time, a data preprocessing method considering the state of three-phase traffic flow is proposed based on the clustering algorithm. The research shows that the aggregation-discrete state of the time series network can reflect the traffic flow state of the road section; the clustering coefficient and the betweenness of the network nodes show an obvious negative correlation. In addition, the key intermediary nodes of different clusters in the network are mainly distributed in the synchronous flow phase of the three-phase traffic flow. The research conclusions will help the traffic management department to take more precise traffic management and control measures.
