Short-Term Traffic Flow Prediction Based on VMD-SMA-SVR

Accurate prediction of short-term highway traffic flow is crucial for ensuring traffic safety and reducing the accident rate. This study proposes a traffic flow prediction model based on Variational Mode Decomposition-Slime Mould Algorithm-Support Vector Regression (VMD-SMA-SVR). The VMD algorithm is used to decompose traffic flow data, and the SMA algorithm optimizes the parameters of the SVR model. The SVR prediction models for each component are constructed and their predictions are aggregated. Comparative experiments with various models demonstrate that the VMD-SMA-SVR model excels in prediction accuracy and stability. The VMD algorithm shows good adaptability, while the SMA algorithm exhibits strong optimization capabilities. This model effectively improves the accuracy of short-term traffic flow prediction, featuring fewer model parameters, high computational efficiency, and low resource consumption, making it suitable for practical short-term traffic flow forecasting.