Моделирование локальных транспортных потоков в условиях неполноты исходных данных

Abstract

The main scientific result of the work is a model implemented in the form of a software package capable of automatically modeling traffic without human intervention, processing data coming from intelligent cameras installed at intersections. Within the framework of the study, the possibility of modeling traffic flows in conditions of incompleteness of the initial data has been confirmed, and specifically, the possibility of modeling flows at intersections, focusing solely on information about traffic in related areas. The study confirms that traffic modeling based on limited data in the future can become a tool of traffic management. The mean squared error of modeling under conditions of incomplete initial data is 3,66, the correlation coefficient between real and simulated data is 0,7, which corresponds to the average degree of coincidence of the results. It is appropriate to emphasize that, when modeling a particular intersection, the correlation coefficient has reached 0,996, which indicates a significant accuracy in modeling this intersection. In two experiments, the correlation coefficients are in the ranges of confidence intervals, which also indicates sufficient convergence of the simulation results. The results of the study are of practical importance with the prospect of determining the optimal parameters of traffic flow control on the model at local time intervals of approximately several tens of minutes. The development of the model is planned in the direction of its use to create adaptive algorithms for controlling traffic light objects at intersections. This step will improve the efficiency of transport infrastructure management to changing traffic conditions.