The insufficient development of intelligent dynamic monitoring systems, which operate with big data, obstructs the control of traffic-related air pollution in regulated urban road networks. The study introduces mathematical models and presents a practical comparative assessment of pollutant emissions at urban intersections, with two typical modes of vehicle traffic combined, i.e., freely passing an intersection when the green signal appears and uniformly accelerated passing after a full stop at the stop line. Input data on vehicle traffic at regulated intersections were collected using real-time processing of video streams by Faster R-CNN neural network. Calculation models for different traffic flow patterns at a regulated intersection for dynamic monitoring of pollutant emissions were obtained. Statistical analysis showed a good grouping of intersections into single-type clusters and factor reduction of initial variables. Analysis will further allow us to control and minimize traffic-related emissions in urban road networks. A comparative analysis of pollutant emissions in relation to the basic speed of passing at the intersection of 30 km/h was performed according to the calculations of the mathematical models. When reducing the speed to 10 km/h (similar to a traffic jam), the amount of emissions increases 3.6 times over, and when increasing the speed to 50 km/h, the amount of emissions decreases by 2.3 times. Fuzzy logic methods allow us to make a comparative prediction of the amount of emissions when changing both the speed of traffic and the capacity of the intersection lanes. The study reveals the benefits of using a real-life measurement approach and provides the foundation for continuous monitoring and emission forecasting to control urban air quality and reduce congestion in the road network.

No abstract

No abstract