Traffic Data Prediction Based on Complex-Valued S-System Model

摘要

To predict traffic data accurately could make an important role in network management. In order to improve forecasting accuracy, this paper proposes complex-valued S-system model (CVSS) forecast small-scale traffic data. According to the form of CVSS model, complex-valued restricted gene expression programming (CVRGEP) is utilized to search the optimal the representation of CVSS. Complex-valued differential evolution (CVDE) is proposed to evolve the parameters of model. The small-scale traffic data is utilized to test our method. Our method has better prediction performances than neural network (NN), radial basis function neural network (RBF), flexible neural tree (FNT), ordinary differential equation (ODE) and S-system.