Traffic Signal Control at an Intersection with Due Regard for the Spatial and Temporal Non-Uniformity of Traffic Flows

This paper discusses the spatial and temporal unevenness of traffic flows on a city’s road network. It demonstrates the uneven distribution of vehicle traffic throughout the day, both at the intersection as a whole and in individual directions, using a key intersection in Tyumen as an example. Machine learning methods are used to cluster vehicle traffic flow rate values based on travel directions. The results reveal four distinct classes, excluding nighttime, each of which requires the development of a separate traffic signal plan. Cluster boundaries are not synchronous with the beginning of each hour, and practical implementation requires the configuration of new traffic monitoring system database templates. Adjusting the timing of signal plan switches to account for hourly unevenness in traffic flow rates can reduce vehicle delays by up to 6.3% during certain periods. The results of the study of traffic unevenness can be used to select methods for improving the efficiency of traffic management at intersections by adjusting signal plans, the order of vehicle movements by direction, and the number of traffic light phases. Traffic unevenness analysis is essential for improving integrated traffic management plans and developing plans to reduce road accidents. The methods for analyzing traffic flow irregularity discussed in the article can be integrated into the city’s digital twin and the Intelligent Transport System.