Traffic signal segmentation algorithm based on twodimensional clustering of traffic volume and vector angles

Clustering analysis is usually used to extract the time points in order to give a reasonable divisions of intersection traffic signal time intervals. However, in consideration that the volume is similar while the direction is different or similar cycle while different green ratio, the timing of the signal will be divided into different time periods. In this paper, a two-dimensional clustering analysis method based on total volume and vector angle is proposed. The traffic volume of intersection is defined as the total volume and the vector angle, two parameters, through clustering analysis of the total volume and the vector angle; we get a collection of two time periods at the end of merging times; by using the merging rules repeatedly, the time is divided into several modes. The two-dimensional analysis algorithm can distinguish the difference of each sub volume under the condition that the total volume is similar; the traffic situation of each mode is similar, and finally matches the appropriate timing scheme. Simulations of Synchro 7 show that compared with the traditional method, the total parking time in the traffic pressure of heavy intersection decreased by about 20%~30%, and the total parking time in the T-shaped intersection decreased by around 15.6%; at the same time, the total parking time get a better decline in intersection with high intensity traffic pressure through the field detection.