Automatic Identification of Near Stationary Traffic States Using Changepoint Detection Method

• Published in: Transportation research record Vol. 2676; no. 9; pp. 187 - 201
• Main Authors: Chen, Wangyong, Hu, Yao, Hu, Qian, Shen, Qi
• Format: Journal Article
• Language: English
• Published: Los Angeles, CA SAGE Publications 01-09-2022
• Subjects: congestion data and detection; data and data science; urban transportation data and information systems; analysis; data cleaning; statistical methods
• ISSN: 0361-1981; 2169-4052
• DOI: 10.1177/03611981221086643

Near stationary traffic states are of great significance for the calibration of the fundamental diagram and the quantification of capacity variation. In this paper, based on wavelet transform and robust functional pruning optimal partitioning (RFPOP) changepoint detection, a robust and efficient method for automatic identification of the near stationary traffic states is proposed. This method first removes the noise influence of traffic flow series, divides the series automatically into multiple candidate intervals that may be close to stationary states according to the RFPOP changepoint detection method, and calculates the candidate interval characteristics. The near stationary states are then identified based on the modified Cassidy’s criterion. A case study is provided for the proposed method, and its robustness is proved in a simulation experiment. Finally, it is shown that the method of automatic identification of near stationary traffic states proposed in this paper is robust and effective.