Application of the Neural Network Theory to the Flood Prediction

Flood evolution exhibits a complicated non-linear dynamical process.The neural network possesses the capability of dealing with complexnon-linear dynamical systems,this paper demonstates howit can be used in flood prediction as a new approach considering the non-linear relation ship between flood evolution and its factors such as discharge,channel deformation,and so on.Based on it,the neural network approach is applied to the flow prediction of Yangtze Riverat Luoshan station.The preliminary results suggest that the phenomenon of small discharge with high level in middle reaches of Yangtse River recently,especially in 1998,is related to the downstream aggregation.And the quantitative relations between the water level variation of Luoshan station and the downstream aggregation are obtained.