Threshold Division of Urban Road Network Traffic State Based on Macroscopic Fundamental Diagram and K-Means Clustering

摘要

In order to identify the real-time traffic state of the regional road network effectively, this paper proposes a method to determine the thresholds of the selected index corresponding to different traffic states, which combines the macroscopic fundamental diagram (MFD) of the road network with the K-means clustering. Firstly, in order to facilitate real-time control of regional traffic, the number of accumulated vehicles inside the road network is selected as the discriminant index, and the cluster K value is determined according to the data characteristics of MFD and related specifications. Then select the K-means algorithm to cluster the MFD scatter plots to get the range of indicators corresponding to each traffic state. The case study shows that the method can quickly identify the macroscopic traffic state of the road network according to the number of internal vehicles in the road network. Moreover, the difference between the threshold of the indicator obtained by clustering the MFD and the threshold obtained according to the speed of the road network is within 10 veh, which proves the effectiveness of the method. In addition, by comparing and analyzing the index thresholds identified under different control methods, the application value of this method in the evaluation of traffic measures is illustrated.