Prediction of short-term transportation flow based on optimizing wavelet neural network by Genetic Algorithm

The weights and the parameter of Wavelet basis of the Wavelet neural network function are always initialized randomly, so the evolution of network tends to be local optima and each forecast results will vary widely. Genetic algorithm is used to optimal the weights and the parameter of Wavelet basis function of the Wavelet neural network, to construct a Wavelet neural network which is on the basis of genetic algorithm. In this paper, we apply this method to forecast short-term time traffic flow, verify with instances, and compare with Wavelet Neural Network Method. The results indicates that this method is not only more stable, but more precise.