Consistency Analysis of Traffic State Clustering Identification on a City Expressway

As the result of traffic state clustering, only the characteristics of the data itself are represented; it lacks traffic theory basis. The consistency verification is proposed in this paper. Firstly, using the traffic state dividing theory from the perspective of combining theory and practice, the traffic state dividing machine is constructed based on Fuzzy C-means Clustering. Secondly, the completeness of detecting data on the road section is classified into three types:(1) contains congestion, but lacks top data in the statistical period; (2) contains both congestion and top data; and (3) does not contain congestion. Thirdly, the consistency of the clustering result and traffic state dividing theory is verified by a non-parametric statistical test method. Finally, data is taken from 43 dual-loops on the eastern side of the Shanghai north-south expressway for 10 days as examples, and the empirical analysis is realized. The results show that the samples of the first and second types are all successfully verified. That is to say, the traffic state clustering results of these two types can be taken as the final result of traffic state identification, but the clustering results of the third type need be modified by regression analysis.