Road network traffic state discrimination method based on clustering and graph convolutional network

Abstract

The invention discloses a road network traffic state discrimination method based on k-means + + clustering and a graph convolutional network, and belongs to the technical field of traffic state discrimination. The method includes: firstly, dividing a target road network into n road sections, and obtaining the average flow of each road section at regular intervals; secondly, constructing topological structure data of a road network; clustering the characteristic matrix reflecting the road network traffic state by using a k-means + + clustering algorithm to obtain sample data with labels; constructing a road network traffic state discrimination decision model based on a graph convolutional neural network; and finally, dividing the sample data into a training sample set and a test sample set, training the road network traffic state discrimination decision model by using the training sample set, and testing the accuracy of the model by using the test sample set. The road network serves as a research object, the traffic state of the whole road network is researched, the traffic state judgment decision model based on the graph convolutional network is provided, the model can fully consider the spatial characteristics of the road network, and the real-time performance and accuracy of road network traffic state judgment are improved.

Description

Technical field [0001] The present invention relates to a road network traffic state discrimination method, which belongs to the field of traffic status. [0002] technical background [0003] With the continuous improvement of my country's economic level, the number of motor vehicles in the city has a sharp rise, resulting in a significant increase in traffic flow, compared with the construction of urban roads in my country, is relatively slow. Under both contradictions, traffic congestion has become a common problem in the city, which seriously affects the development of urban economy and the life experience of residents. In order to alleviate traffic congestion, urban traffic needs to be scientific and fine management. Research on the route traffic status discrimination method can make traffic managers better understand the overall traffic state of the road network, there is an important means of transportation management measures, which is important in traffic control, transpo...

AI Technical Summary

Problems solved by technology

Commonly used machine learning algorithms include k-nearest neighbors, support vector machines, and deep learning. Compared with k-nearest neighbors and support vector machines, deep learning has stronger feature extraction capabilities. However, the current traffic status based on deep learning Most of the discriminant methods lack the consideration of the spatio-temporal characteristics of the road network

[0005] In addition, at this stage, most of the research on road network traffic status based on machine learning methods is carried out on the research of road segment traffic status, ignoring that different road segments have different effects on the overall traffic status of the road network

Images