In order to overcome the drawbacks in linear combination forecasting modeling, the wavelet network method was applied to establish the optimal combination function for developing the nonlinear combination forecasting modeling of urban water consumption. By this method, the fitting of the nonlinear combination function was transformed into parameters estimation of wavelet network. The forgetting factor method was used to train the weights of wavelet network and scale factors, and displacement factors were trained by the predicting error method. The numerical example shows that the proposed comb...