An abnormal traffic state detection method based on dynamic hypergraph convolutional neural network

Abstract

The application discloses an abnormal traffic state detection method based on a dynamic hypergraph convolutional neural network, which is suitable for the field of traffic big data, and enhances the abnormal detection capability of the model for unbalanced abnormal label traffic data and improves the robustness. In order to increase the robustness and effectiveness of the model, a data enhancement technology is introduced, the original data is twisted and masked in the time and space dimensions, so that the model can better detect abnormal conditions from these unbalanced abnormal data. Secondly, a spatio-temporal hypergraph convolution block is designed to capture the time characteristics and spatial characteristics of traffic data. A dynamic evaluation mechanism is also designed to represent the evolution attribute of the road mode, so as to more accurately describe the dependence relationship between different roads. Finally, a dynamic hypergraph convolutional neural network is proposed for detecting abnormal conditions in complex traffic scenes. The method provides an effective method for solving the problem of traffic abnormal state detection.

Images