Abstract: | Flood propagation exhibits a complicated non-linear dynamical process. An artificial neural network (ANN), capable dealing with complex non-linear dynamical systems, is used for flood prediction in this paper. The ANN considers the non-linear relationship between flood evolution and effective factors such as discharge and channel deformation. The ANN is applied to the flow prediction of Yangtze River at Luoshan station. The preliminary results suggest that phenomenon of low discharge but high stage in the middle Yangtze River in 1998 is related to the downstream aggregation. And quantitative relations between water stage variation at Luoshan station and downstream aggregation are obtained. |