Graph neural network traffic flow prediction method based on multivariate time sequence interpolation

Abstract

The invention discloses a graph neural network traffic flow prediction method based on multivariate time sequence interpolation, and the method comprises the steps: describing the topology between stations in a traffic network through a weighted graph, and enabling the inflow and outflow of traffic to serve as the characteristics of station nodes; constructing a traffic flow graph according to the inflow and outflow conditions of each station; interpolating the missing value by using a graph neural network; obtaining a spatial feature sequence through a graph attention network; capturing the dynamic correlation between different times through time attention, and then capturing the time characteristics by using the long and short-term memory network to obtain the traffic flow characteristics as the prediction result. The method can eliminate the influence caused by the missing of the time-space relation data in the traffic flow prediction, overcomes the data missing possibly occurring in the practical application, and achieves the long-term prediction of the traffic flow.

Description

technical field [0001] The invention belongs to the technical field of traffic flow prediction, in particular to a graph neural network traffic flow prediction method based on multivariate time series interpolation. Background technique [0002] With the rapid development of various positioning technologies such as the Global Positioning System (GPS) and mobile devices, and the continuous improvement of information collection equipment such as cameras and sensors, a large amount of traffic time series data with geographic information has been accumulated for data analysis. We have entered the "big data era", which has had a major impact on human life and social governance. Therefore, how to obtain useful information from massive data has attracted widespread attention. Spatio-temporal data mining plays an important role in many applications, including intelligent transportation, urban planning, public safety, healthcare, and environmental management. Traditional data mining...

AI Technical Summary

Problems solved by technology

In practical applications, the data collected by some institutions and organizations is not real-time, and sometimes the data will be lost, which will have a great impact on traffic flow forecasting, and using these data to predict traffic flow will obviously cause large deviations

Images