G.C. Ao, H.W. Chen, H.L. Zhang

Pages: 23-30

Abstract
Changes in the expressway traffic state are in essence the temporal and spatial variations of traffic flow parameters. According to the real traffic flow data of a three-lane road section of an expressway in Beijing, there is a difference in characteristic between the traffic flow in the cross section and the traffic flow in lanes. By comparing the occupancy-volume data with traditional two-phase traffic flow pattern map and three-phase traffic flow pattern map, we can see that the traditional traffic phase theory is hard to describe the state of urban expressways and the occupancy-volume data are universally discrete. We regard the occupancy-volume standard deviation as a non-stationary time series and process it with BG algorithm, and then we obtain the change point of data discretization and based on this, classify the traffic flow of expressways into unblocked flow, congested flow and jam flow. The measured occupancy-volume data indicates that the classification of traffic state based on the change point coincides with the actual operation status of the vehicle and that it is well adaptable under all road conditions with well-equipped traffic flow detection equipment.

Keywords: urban traffic; traffic state; traffic characteristic analysis; urban expressway; traffic status classification