Research on Traffic Prediction Model Based on KPCA

摘要

In order to optimize communication network architecture and analysis network capacity, in this paper a traffic prediction model based on kernel function principal component analysis(KPCA) is presented in view of the traffic features. In this model KPCA was used to extract nonlinear features of input data. Those nonlinear features can reflect the relationship among the input data better than the method of principal component analysis (PCA). By the way dimension of input data of neural network are reduced, sensitivity of traffic change is improved by input data and the prediction model was simplified. Simulation results show that prediction model based on KPCA-RBFNN has generalization performance and good ability to deal with nonlinear data.