Traffic prediction method and device based on dynamic space-time diagram convolution attention model

Abstract

The invention discloses a traffic prediction method and device based on a dynamic space-time diagram convolution attention model. Comprising the following steps: modeling a road traffic network into a graph structure; modeling a dynamic space-time diagram convolution attention model by using the diagram structure and the diagram signal; through an encoder unit of the dynamic space-time diagram convolution attention model, capturing dynamic space correlation among sensor nodes at different moments, capturing time correlation of step lengths at historical moments and time correlation among multi-step prediction, and obtaining a hidden state including each prediction step length; and enabling the decoder unit to receive the hidden state of each prediction step length, and obtaining a prediction result of recording speed values of all sensor nodes in the next one hour by adopting a scheduling sampling strategy. Experiments prove that the method of the invention can effectively obtain a traffic prediction result in the next one hour, and the model of the method of the invention exceeds SOTA.

Description

technical field [0001] The invention belongs to the field of intelligent transportation, and in particular relates to a traffic prediction method and device based on a dynamic space-time graph convolution attention model. Background technique [0002] Traffic status prediction is a key task in the field of Intelligent Transportation System (ITS). The real-time and accurate prediction of traffic flow and speed plays an important role in relieving traffic pressure, coordinating traffic management and building smart cities. Traffic data comes from sensors installed on the edge of the road, and the recording period is usually 5 minutes. Due to the connectivity of the road and the inherent serial correlation of data records, traffic data is a typical spatio-temporal data. Early traffic state prediction methods were mainly model-driven, such as methods based on linear regression, autoregressive algorithms based on sliding windows, and Kalman filter algorithms. However, these algo...

AI Technical Summary

Problems solved by technology

Therefore, the models proposed by previous researchers are not enough to model the dynamic spatial correlation. Although the authors of DSTGCN have added the tensor decomposition operation into the end-to-end learning framework, the tensor real-time decomposition will inevitably bring higher Computational complexity, and the model still relies on a predefined adjacency matrix

Images