Abstract: On the basis of theoretical analysis of the early water supply warning mechanism, gathering a variety of information contained with some single models, the best way combination with single models are proposed based on the results of early warning mechanism based on the early warning of optimized combination with neural network, it can forecast error with gray model results. The model has good predictive and early warning effect, so as to provide a new water supply early warning methods and ideas.

Key words: water warning, neural network, gray model, combination forecasting model