Modeling Traffic Flows Using Neural Networks

The object of the study is the process of traffic flows on the urban street and road network of a metropolis. Due to increasing urbanization, its negative consequences are also increasing, including a decrease in the speed of communication, leading to the formation of traffic congestion. The aim of the work is to develop a mathematical apparatus for controlling congested traffic flows on the urban road network using neural networks. The results of the study are aimed at solving the problem of traffic congestion in cities, especially in the largest ones. The methodological basis of the research includes a systematic approach, systems analysis, synthesis, theories of traffic flows, systems, dynamic systems, complex systems and chaos. The objective function of the study is to maximize the kinetic energy of the transport flow. As an optimization criterion, it is proposed to use a measure of traffic chaos – the entropy of traffic flows. According to the results of the study, a mathematical model was obtained for changing the kinetic energy of the traffic flow under the influence of the relative entropy of the lane occupancy by traffic at controlled intersections. It is proposed to use street (road) video surveillance cameras to monitor traffic flows. For this purpose, mathematical models have been developed that allow measuring the occupancy of traffic lanes at specific controlled intersections by processing video images with neural networks in real time. Experimental studies are conducted to confirm the developed mathematical apparatus. The article presents the current results of the experiment, which confirm the proposed mathematical models. The obtained research results are valid only taking into account the specified restrictions and requirements for traffic flow. On the one hand, this significantly narrows the possibility of applying the results in practice, on the other hand, it allows to increase the accuracy and purity of experimental research.