INCREASING THE THROUGHPUT AT REGULATED INTERSECTIONS BY OPTIMIZING THE SPEED MODES OF TRAFFIC FLOWS
Authors
Olga Fadina
South Ural State University
Vladimir Shepelev
South Ural State University
Mikhail Varvorkin
South Ural State University
Lev Pluhin
South Ural State University
Keywords:
neural networks, intersection throughput, saturation flow, speed, intelligent transport systems, queue of vehicles, coordinated traffic flow management, traffic flow, traffic, intersection, time to leave the queue of vehicles
Abstract
This study presents a detailed analysis of the traffic flow parameters at regulated intersections using machine vision. Based on the processing of video streams by a trained and optimized neural network (YOLOv4), data analysis was carried out to assess the capacity of lanes with traffic permitted only in a straight line, the characteristics of intersections were collected, and a mathematical model was developed for calculating the average speed of group cars to ensure non-stop passage of a regulated intersection with coordinated traffic management. The dependences of the average speeds of the leading car on the time of departure from the queue of vehicles were obtained. The proposed method makes it possible to increase the throughput of regulated intersections by up to 12 % and reduce the delay time of vehicles by up to 20 %.
Author Biographies
Olga Fadina, South Ural State University
postgraduate student, assistant at the Department of Automotive Engineering
Vladimir Shepelev, South Ural State University
Associate Professor of the Department of Automotive Engineering
Mikhail Varvorkin, South Ural State University
student of the Department of Automotive Engineering
Lev Pluhin, South Ural State University
student of the Department of Automotive Engineering
