Nonlinear combination of travel-time prediction model based on wavelet network

Abstract

In the paper, research is focused on a combination of artificial neural network and Kalman filtering theory with application to real-time travel-time prediction model. ANN forecasters and Kalman filtering can model the complicated relationship between travel-time and traffic volume in related links. To enhance the prediction accuracy of these models, a nonlinear combination prediction approach of these two models is proposed based on wavelet networks. The performance of the novel model is tested by real detected traffic data or the links in the urban road networks. The results indicate that combination strategies based on the wavelet network outperform the other approaches.