Interval data‐based k ‐means clustering method for traffic state identification at urban intersections

Abstract

Identifying traffic states at urban intersections plays a significant role in achieving the full potential of intelligent transportation systems for various traffic applications (e.g. real-time traffic signal control). Many traffic state identification methods use mean values of measured or estimated traffic flow variables to identify traffic states; however, lacking a consideration of traffic flow uncertainty leads to inaccurate identifications. To investigate the impact of traffic flow uncertainty on intersection traffic states, this article proposes an interval data-based k -means clustering method for traffic state identification at urban intersections. The uncertainty of three traffic flow variables (volume to capacity ratio, queue length, and delay) are represented in the form of interval data and employed as input variables. The proposed method was implemented on a real-world traffic network in Kunshan, China. Test results show that the clustering results are explicable and can accurately describe the trend of traffic state evolution. Further investigation shows that the proposed method outperforms the mean-value-based method, and queue length has a more significant contribution to the clustering results after the use of interval data. The findings of this study demonstrate the effectiveness of the proposed method in traffic states identification at urban intersections.