Research on Intersection Traffic Volume Prediction Based on VMD-DBO-LSTM

Establishing an accurate and effective intersection traffic flow prediction model can improve the efficiency of intersection traffic management and provide scientific support for traffic policy formulation. This paper studies the problem of intersection traffic flow prediction. First, the data quality is ensured by performing preprocessing operations such as anomaly detection and missing value filling on the original data; secondly, based on the traditional LSTM model, the variational mode decomposition (VMD) method is introduced to extract multi-scale features in the data to capture regular information of different frequencies; then, the key parameters of VMD are optimized in combination with the dung beetle optimization algorithm (DBO) to further improve the adaptability and prediction accuracy of the model. Thus, a VMD-DBO-LSTM hybrid model is constructed to make up for the shortcomings of traditional methods in dealing with nonlinear and non-stationary traffic data. Finally, the actual traffic flow data of a certain intersection is used for example analysis and compared with the benchmark model for verification. The results show that the proposed hybrid model shows better performance in predicting traffic flow at vehicle-road cooperative intersections.