METHOD FOR PREDICTING TRAFFIC DYNAMICS IN A ROAD SYSTEM Russian patent published in 2021 - IPC G08G1/01 

FIELD: road systems.

SUBSTANCE: invention relates to a method for predicting the dynamics of road traffic in a road system. A method for predicting traffic dynamics in a road system contains several stages. Traffic profiles are combined with their tags into at least two clusters based on the similarity of traffic profiles. At least one characteristic vector is assigned to each cluster based on tags in the specified cluster. Each characteristic vector contains a tag or a combination of tags of the specified cluster. It is determined which cluster contains the largest number of traffic profiles. The specified cluster is set as a reference cluster. The measure of dissimilarity of the specified cluster with the reference cluster is determined for each cluster. The specified measure of dissimilarity is assigned to all the characteristic vectors of the specified cluster. The measure of dissimilarity is a measure of how much the traffic profiles of the cluster differ from the traffic profiles of the reference cluster. A request is accepted containing one or more request tags. It is checked whether all the request tags correspond to one of the characteristic vectors of the clusters. If they correspond, a prediction of traffic dynamics is issued, based on a cluster with the specified corresponding characteristic vector. Otherwise, one of the request tags is excluded from the request and a return to the beginning of the specified validation step is made. The request tag to be excluded is selected by determining which characteristic vectors correspond to each request tag, and the request tag is selected, for which the corresponding characteristic vector has the smallest measure of dissimilarity.

EFFECT: improved road safety is achieved.

10 cl, 6 dwg