A method of repairing traffic road network data based on graphsage-gan

Abstract

A GraphSAGE-GAN-based traffic road network data restoration method, first calculates the correlation between road network detectors according to the historical data of the detectors in the road network, obtains the road network correlation matrix, and then calculates the correlation matrix according to the obtained road network correlation The matrix constructs the road network structure based on time correlation. Secondly, use GraphSAGE to extract the potential spatio-temporal features of the road network structure, and finally use the extracted spatio-temporal features as the input of the generator in the generative adversarial network, so that through the adversarial training of the generative adversarial network, the generator can use the spatio-temporal features extracted by GraphSAGE Generate complete road network traffic status information, so as to realize the restoration of road network traffic status data. The invention can deeply excavate the spatial characteristics between road network traffic state detectors, and effectively improve the accuracy of road network traffic state restoration.

Description

technical field [0001] The invention relates to a method for repairing traffic road network data based on GraphSAGE-GAN, which belongs to the field of intelligent traffic. Background technique [0002] In the process of building a smart traffic city, road traffic status data plays a central role. Complete data can better enable road managers to obtain accurate road network traffic status information, thereby better regulating the operation status of the road network. However, in reality, the road detector's own failure, and various irresistible factors (such as: earthquakes, floods caused by power failure of the road traffic state detector, etc.) cause the traffic state information acquired by the road detector to be incomplete. Therefore, for road network traffic status data, it is of vital significance to use data restoration algorithms to interpolate missing data and ensure the integrity of road network traffic status data. [0003] The current road traffic prediction me...

AI Technical Summary

Problems solved by technology

[0003] The current road traffic prediction methods mainly include: historical average method, K nearest neighbor method, noise reduction self-encoder algorithm, generative confrontation network interpolation algorithm, etc.; existing technical defects: unable to deeply mine the gap between road network traffic state detectors Spatial features, the accuracy of road network traffic status restoration is low

Images