River Confluence Forecasting Model Based On Artificial Neural Networks And Applied Research

To rivers which have several branches, there is close water power relation between main current and branches, the runoff of main current has close relation with each branch which have water volume contributes to main current. Main current basin is often the relative developed area. The runoff of main current affects development of main current basin's social economy directly, and influences water environment condition remarkable.Therefore, obtains water volume of main current can raises the space and time distribution of water resources for water volume dispatch, water resources plan and the management,water resources conservation.It has important theory significance and practical significance to reseach the model of water volumn forcast.In view of concentration calculation problem, this article involvs the multi-branches rivers concentration calculation problem to conduct the research.Firstly, based on reading literatures massively, analyzes present situation of concentration calculation research in domestic and foreign, and points out the research prospect of basin concentration calculation in the future; Secondly,summarizes present situation of the Artificial Neural Network in the hydrology and water resources domain, and forecasts the research prospect of ANN in the hydrology and water resources domain; Thirdly, introduces ANN and the gray connection analysis method; Fourthly, studies concentration process of multi-branches rivers and its hydrology characteristics; Fifthly, proposes a new concentration calculation method which is suitable to multi-branches rivers, namely gray-grtificial neural network concentration calculation method; Finally, applies the method in Tarim River's three sources and courses concentration calculation, has confirmed the method is feasibility, and rationality.The core contents of the article are:(1)Proposes the gray-artificial neural network concentration calculation method which is suitable to multi-branches river. This method carries on according to two steps, firstly, uses gray speed interconnect degree to calculate the concentration time of various sources to main current.Secondly, choose the influence factors and establishes ANN model, after trains the network, we can obtain the concentration calculation model. (2) Applies GACC to carry on three sources and courses concentration calculation in Tarim River.Take Tarim River's three big branches: Akesu River, Yeerqiang River and Hetian River hydrologic station's discharge of water and sum of pilots of various sources as input factors, take the convergence place Ala'er hydrologic station's discharge as the output factor, establishes ANN model and trains it.Finally we can obtain the Tarim River's three sources and courses affluxes forecast model.Through the study to Tarim River, indicates this model is reasonable and effective. The model can provides water volume dispatch basis for water resources management and so on.