Expressway traffic flow forecasting based on a combined model

Accurate expressway traffic flow forecasting can provide reliable traffic flow forecasting data for traffic management departments and help them dispatch traffic resources rationally. However, the single model cannot accurately forecast complex traffic flow data. To improve the forecasting accuracy of expressway traffic flow, this paper proposes a combined forecasting model. The model not only uses principal component analysis (PCA) and a variety of data preprocessing methods to perform feature processing on original data, and selects four optimal models as the sub-models of the combined model based on the testing accuracy of six different models, but also uses the seagull optimization algorithm (SOA) and the grey wolf optimizer (GWO) to optimize the weights of the combined model, and by comparing the forecasting accuracy of the combined models to select an appropriate optimization algorithm for the combined model in this paper. The experiment results indicate that the proposed combined forecasting model obtains high accuracy and stability.