Short-term traffic flow prediction of expressways based on VMD-XGBoost

To improve the accuracy and robustness of short-term traffic flow prediction on expressways, and address the periodic and fluctuating components in traffic flow time-series signals, this paper proposes a combined VMD-XGBoost prediction method. First, the method uses Variational Mode Decomposition (VMD) to adaptively decompose the original traffic flow signal into low-frequency periodic components and high-frequency fluctuating components. Then, aiming at the characteristics of different components, XGBoost models are constructed respectively and trained in a targeted manner. Finally, the prediction results of each sub-model are fused to obtain the final prediction result. A case study is conducted using the PeMS dataset, and the results show that compared with the single XGBoost model, the VMD-XGBoost model considering signal modal decomposition can significantly improve the prediction accuracy of short-term traffic flow. This model can effectively mine the time-series characteristics of traffic flow and provide decision support for road traffic management and other work.