Experiment and Simulation of Flood Increase along Lower Yellow River

In view of the abnormal phenomenon that a flood peak increased in August 2004 , July 2005 and August 2006 along the lower Yellow River, the experiments of this abnormal phenomenon is studied. It is found that the flood increase was due to the decrease the channel roughness in the propagation of high concentrated flood carrying the extra fine sediment which was discharged from xiaolangdi reservoir. Experiments with hyper-concentration flowing over the rough beds prove that the resistance of the flow may be considerably reduced by suspended sediment. Especially in hyper-concentration, the phenomenon of dragreducing is more obvious when the sediment contents exceed the flocculation sediment contents. Based on the experiment result, a chaotic BP neural network model is proposed in this paper. Based on the chaos identification to the flood system, chaos BP neural network model are developed combined chaos theory and BP neural netwok, flood sequences are disposed by phase-space reconstruction to be as training sample. Network structure can be determined by Matlab toolbox. The established chaos BP model is used to predict the phenomenon of peak value for Huayuankou hydrometric station in 2006. The results show that the predictive model combined chaos theory and BP neural network, has certain reference value to improve flood forecasting accuracy as a new attempt.