Discussion of 'A Wavelet Network Model for Short-Term Traffic Volume Forecasting' by Yuanchang Xie and Yunlong Zhang

摘要

This paper presents a Wavelet Network (WN) model for forecasting short-term traffic flow. The Levenberg-Marquardt learning algorithm is employed to train the WN model. Earlier, Jiang and Adeli (2005a) demonstrated that the Wavelet Neural Network (WNN) model outperformed the conventional neural network in traffic flow forecasting and elucidated several important issues regarding the WNN structure and wavelet type. First, the advantages of developing the WNN model for traffic flow forecasting need to be clarified. It is well known that the simple backpropagation neural network model has its inherent shortcomings such as lack of an efficient constructive model, slow convergence rate resulting in excessive computation time, and entrapment in a local minimum (Adeli, 2001). In contrast, the WNN model uses a wavelet function with a spatial-spectral zooming property, which influences the output of the model only in the finite range of input data. This property has two advantages for traffic flow forecasting: (1) it reduces the undesirable interaction effects among the nodes of the neural network, thus in general improving the accuracy of traffic forecasting, and (2) it accelerates the neural network training process, thus improving its computational efficiency in traffic forecasting.