Short-term traffic flow prediction based on Variational Mode Decomposition and Gated Recurrent Unit

With the increase of urban population and the number of cars, the importance of traffic flow prediction has become increasingly prominent. The traffic flow data is nonlinear, periodic and random. The unstable traffic flow data depends on the long-term data range, and the original data will contain some noise due to some external factors, which may lead to the decline of the prediction accuracy. To solve the above problems, this paper introduces Variational Mode Decomposition (VMD) for noise reduction processing of vehicle flow data. Then, the Gated Recurrent Unit (GRU) network that incorporates the Attention mechanism (ATT) solves the problem of long-term data dependence. Furthermore, Sparrow Search Algorithm (SSA) is introduced to optimize the hyperparameters of GRU, and the VMD-SSA-GRU-ATT short-term traffic flow prediction model is proposed. This paper takes PeMS and M26 traffic flow datasets as the example to make prediction analysis. The simulation results show that the proposed method has higher prediction accuracy and smaller prediction error.