Time delay neural network application to flood forecast
Jinglin Qian · Journal of Hydroelectric Engineering · 2009
This paper presents a method of delay neutral network for simulation of the rainfall runoff process.By applying the Takens phase space reconstruction theory,the rainfall time series of previous period is reconstructed,and the vector obtained is then used as the node input of the neutral network.This method is able to eliminate the subjectivity of existing methods in input estimation,and provides a theoretical basis to determine the correct values of the neutral network input.A case study shows the high efficiency and good accuracy of this method in rainfall runoff forecasting.