Artificial neural network model for flood water level forecasting
Xingming Zhu · Journal of Hydraulic Engineering · 2005
The artificial neural network technology is applied to establish the model for forecasting the flood water level based on the data of upstream hydrological station and local station. The deterministic coefficient in forecasting norm for hydrology is taken as the objective function. For improving the network training speed and forecasting the probable high water level exceeding the highest water level in history, a standardized method is given to treat the data of input layer and output layer. The proposed forecasting model is applied to analyze the hydrological data of two stations located in Beijiang River, Zhujiang Delta. The result shows that by reasonable selection of original data for input layer element and forecasting period, satisfactory forecast precision can be obtained.